Nvidia released the GeForce 546.34 WHQL driver for Windows 10-64 bit and Windows 11, which is the first to support the RTX 4090D exclusively made for the Chinese market as a way to conform to the export regulations imposed by the US.

Besides supporting the RTX 4090D, Nvidia delivers 'Game Ready' support for The Finals and Squad. Fortnite Chapter 5 gets DLSS Super Resolution support. The 546.34 WHQL driver fixes a specific issue with Discord where colors appear muted when the gameplay is streamed and another issue where a new Nvidia icon is created in the system tray every time a user switch occurs in Windows. There is one open issue where Netflix has display issues via the Edge browser.

The GeForce 546.34 WHQL driver only supports the new GeForce RTX 4090D and can be downloaded from Nvidia's website.

This sanction-proof RTX 4090D was launched two days ago in China with 12.8% fewer CUDA cores than the standard variant to comply with these restrictions mandated by the US government. Chipmakers can only distribute and sell GPUs under 4,800 TPP (Total Processing Power) to ensure the Chinese market does not retail any high-performance hardware and AI accelerators. This prompted Nvidia to make this graphics card since the RTX 4090 is 5,286 TPP. AMD also complies with this sanction restriction by providing a scaled-down AI accelerator. These GPUs are needed in the Chinese retail market and system builders like Dell as the company had to stop bundling its systems with a plethora of systems with AMD Radeon graphics cards.

China is a significant market for respective manufacturers and vendors to ignore; hence, making some changes to comply with these restrictions makes it well worth the effort. We'll likely see more 'under 4,800 TPP' GPUs from respective chipmakers as long as the sanctions are active, either for China or any country with this restriction.

The RTX 4090D, made by some AIC partners, is sold via Chinese-based retailers like JD between ￥13,999 ($1,977.66) and ￥16,999 ($2,401.47). Before this GPU's availability, Chinese retailers and scalpers could have some stop-gap solutions by importing RTX 40 series GPUs from other countries. At the same time, some domestic GPU brands repurposed older graphics cards. This GPU might be available in a few other countries and continue to roll out support via globally released drivers like the rest of the GPUs.

Source

More than a year ago, AMD announced the first entry in the RX 7000 series. The Radeon RX 7900 XTX and RX 7900 XT graphics cards.

The $999 AMD RX 7900 XTX couldn’t take on the $1599 Nvidia GeForce RTX 4090 in performance, to everyone’s disappointment. However, it competed nicely against the $1199 Nvidia RTX 4080. The $899 AMD RX 7900 XT remains a disappointment.

On the other hand, the $499 RX 7800 XT turned out to be a bestseller, thanks to it’s cheaper price and good performance. RX 7700 XT, however, costing $449, didn’t seem to sell that well.

But, what about the budget range. AMD also released the RX 7600 graphics card for $269. So what has happened is that while AMD has a graphics card in the sub $200 and sub $400 range, it has nothing in between in the sub $300 range to fill in that gap.

Looks like that might change.

RX 7600 XT Rumors Leak

A few days ago, VideoCardz reported that AMD could be preparing RX 7600 XT, RX 7700 (non-XT) and RX 7800 (non-XT) graphics cards. However, as per the latest Benchlife report (translated), via VideoCardz, that’s not entirely correct.

Benchlife reveals that as per its sources, AMD is going to launch RX 7600 XT in the later part of January 2024. About two days before the date on which Nvidia releases the RTX 4070 Ti Super graphics card.

Additionally, VideoCardz had reported in September that RX 7600 XT could come in two variants. A 10GB version and a 12GB version.

It must be mentioned, RX 7600 uses the largest Navi 33 XL GPU. It means there’s no place to increase the shader count on that particular GPU. So AMD is all likely to use a cut-down version of the more powerful Navi 32 XL GPU found in the RX 7800 XT and RX 7700 graphics cards.

However, there seems to be no plans of releasing RX 7700 (non-XT) and RX 7800 (non-XT) graphics cards by AMD.

It is also reported that AMD might release next-gen RDNA 4 based graphics cards in the second half of 2024, with PCIe 5.0 support. However, it could be postponed to 2025 too.

The report concludes by saying that AMD might not release any reference Made By AMD graphics card for the RX 7600 XT and will stick to the aftermarket sellers like ASUS, MSI and others.

Update: New post by @harukaze5719 shows that AMD could be preparing RX 7600 XT with 16GB of VRAM. Nothing can be said to be confirmed, though.

Source

According to a new report, Microsoft is busy preparing the next-gen Surface Pro and Laptop, with the announcement scheduled for the first half of 2024 (allegedly in spring). The Surface Pro "10" and Surface Laptop "6" will bring notable all-around changes, not just new chips inside.

Speaking of chips, Microsoft will offer more Surface computers with ARM processors for the first time. Sources within Microsoft claim the upcoming computers will be available in Intel and Qualcomm flavors: 14th Gen Intel "Meteor Lake" and Spandragon X Series. Previously, ARM processors were only available in the Surface Pro lineup.

Both Intel and Qualcomm configurations will be bundled with dedicated neural processing units (NPUs) for accelerating AI-dependent tasks. However, Microsoft will allegedly put more emphasis on the ARM variants as purpose-built devices for the next generation of Windows (codenamed "CADMUS PC"). These CADMUS computers will rival Apple Silicon-based devices with their hard-hitting performance and unrivaled battery life.

The tenth-generation Surface Pro will bring new chips and notable display updates. Look out for better HDR support with higher max brightness, new anti-glare coating, and rounded corners. Interestingly, Microsoft is also experimenting with lower-cost models that feature lower display resolution (down from 2880x1920 to 2160x1440).

Additional changes in the Surface Pro "10" will include NFC support for commercial users, a reworked front-facing camera with a wider lens, new colours, and an updated Type Cover with a dedicated Copilot button.

As for the Surface Laptop "6," expect thinner bezels and rounded corners, two display sizes, more ports (at least two USB-C, one USB-A, and the Surface Connect), and a new haptic touchpad. Besides, the computer will get a dedicated button for invoking Copilot (yay).

The Surface Pro "10" and Surface Laptop "6" should get the most attention in 2024, but Microsoft is also working on a new Surface Laptop Go and a successor to the Surface Laptop Studio 2 (expected in 2025). In addition, the company is experimenting with the idea of expanding the Surface Pro lineup with an 11-inch model and offering the new Surface Pro/Laptop in the old design with refreshed hardware for commercial customers.

Sadly, there is no information on whether Microsoft plans to return to bold form factors and devices like Surface Duo or Neo. With Panos Panay gone, it seems like Microsoft is placing all its Surface money on low-risk, well-established devices.

Are you excited about the next-gen Surface Pro and Surface Laptop? Let us know in the comments.

Source: Windows Central

Source

In many ways, 2023 was a long-awaited return to normalcy for people who build their own gaming and/or workstation PCs. For the entire year, most mainstream components have been available at or a little under their official retail prices, making it possible to build all kinds of PCs at relatively reasonable prices without worrying about restocks or waiting for discounts. It was a welcome continuation of some GPU trends that started in 2022. Nvidia, AMD, and Intel could release a new GPU, and you could consistently buy that GPU for roughly what it was supposed to cost.

 

That's where we get into how frustrating 2023 was for GPU buyers, though. Cards like the GeForce RTX 4090 and Radeon RX 7900 series launched in late 2022 and boosted performance beyond what any last-generation cards could achieve. But 2023's midrange GPU launches were less ambitious. Not only did they offer the performance of a last-generation GPU, but most of them did it for around the same price as the last-gen GPUs whose performance they matched.

The midrange runs in place

Not every midrange GPU launch will get us a GTX 1060—a card roughly 50 percent faster than its immediate predecessor and beat the previous-generation GTX 980 despite costing just a bit over half as much money. But even if your expectations were low, this year's midrange GPU launches have been underwhelming.

 

The worst was probably the GeForce RTX 4060 Ti, which sometimes struggled to beat the card it replaced at around the same price. The 16GB version of the card was particularly maligned since it was $100 more expensive but was only faster than the 8GB version in a handful of games.

 

The regular RTX 4060 was slightly better news, thanks partly to a $30 price drop from where the RTX 3060 started. The performance gains were small, and a drop from 12GB to 8GB of RAM isn't the direction we prefer to see things move, but it was still a slightly faster and more efficient card at around the same price. AMD's Radeon RX 7600, RX 7700 XT, and RX 7800 XT all belong in this same broad category—some improvements, but generally similar performance to previous-generation parts at similar or slightly lower prices. Not an exciting leap for people with aging GPUs who waited out the GPU shortage to get an upgrade.

 

The best midrange card of the generation—and at $600, we're definitely stretching the definition of "midrange"—might be the GeForce RTX 4070, which can generally match or slightly beat the RTX 3080 while using much less power and costing $100 less than the RTX 3080's suggested retail price. That seems like a solid deal once you consider that the RTX 3080 was essentially unavailable at its suggested retail price for most of its life span. But $600 is still a $100 increase from the 2070 and a $220 increase from the 1070, making it tougher to swallow.

 

In all, 2023 wasn't the worst time to buy a $300 GPU; that dubious honor belongs to the depths of 2021, when you'd be lucky to snag a GTX 1650 for that price. But "consistently available, basically competent GPUs" are harder to be thankful for the further we get from the GPU shortage.

Marketing gets more misleading

If you just looked at Nvidia's early performance claims for each of these GPUs, you might think that the RTX 40-series was an exciting jump forward.

 

But these numbers were only possible in games that supported these GPUs' newest software gimmick, DLSS Frame Generation (FG). The original DLSS and DLSS 2 improve performance by upsampling the images generated by your GPU, generating interpolated pixels that make lower-res image into higher-res ones without the blurriness and loss of image quality you'd get from simple upscaling. DLSS FG generates entire frames in between the ones being rendered by your GPU, theoretically providing big frame rate boosts without requiring a powerful GPU.

 

The technology is impressive when it works, and it's been successful enough to spawn hardware-agnostic imitators like the AMD-backed FSR 3 and an alternate implementation from Intel that's still in early stages. But it has notable limitations—mainly, it needs a reasonably high base frame rate to have enough data to generate convincing extra frames, something that these midrange cards may struggle to do. Even when performance is good, it can introduce weird visual artifacts or lose fine detail. The technology isn't available in all games. And DLSS FG also adds a bit of latency, though this can be offset with latency-reducing technologies like Nvidia Reflex.

 

As another tool in the performance-enhancing toolbox, DLSS FG is nice to have. But to put it front-and-center in comparisons with previous-generation graphics cards is, at best, painting an overly rosy picture of what upgraders can actually expect.

Intel plays catch-up

Intel Arc's first year on the market has cemented its status as "a good first try." It could have gone either way based on last year's launch, which was hampered by buggy drivers and inconsistent performance, especially in older games.

 

 

But to its credit, Intel significantly improved its software in the last year, eliminating bugs, fixing annoying problems, and boosting performance in older games. The company has made the most progress on some older DirectX 9 and DirectX 11 games, thanks at least in part to code translation technologies that allow those APIs to run on top of DirectX 12 and/or Vulkan, newer low-level graphics APIs that Arc GPUs are better at handling.

 

Intel has also remained relatively competitive on price, thanks partly to Nvidia and AMD's aforementioned underwhelming midrange GPU launches. The Arc A750 is consistently available for $200 or a bit less, making it a solid value for TK

 

But will Intel stay committed to the GPU market? At this point, the company is still moving ahead with its current road map, which should get us next-gen "Battlemage" GPUs at some point in 2024. Arc technology and branding has also made its way into Intel's latest integrated GPUs.

 

But current Arc GPUs are thoroughly outmatched in performance and power efficiency by Nvidia and AMD's offerings, locking Intel out of the $300-plus GPU market. In the last year, the graphics division has seen some leadership shake-ups, and Intel seems particularly eager to shed unprofitable side hustles (like its crypto-mining chips or the NUC mini desktops) as it tries to improve its financial position. Arc cards have yet to break out of the "other" category in the Steam Hardware Survey (though, to be fair, most of AMD's RX 7000 GPUs haven't either).

 

It's hard to imagine that Intel will keep plowing the resources into developing and marketing new GPUs if it seems like it will be stuck selling a relatively small handful of low-margin midrange and low-end chips. 2023 was a decent year for Arc, but 2024 may determine its future.

Power efficiency improves

Maybe the performance of 2023's new GPUs has been underwhelming, but they have a few things going for them. The biggest is power efficiency—if you're using a newer and more power-efficient manufacturing process, and you're not using the extra headroom to boost performance, then you'll mainly benefit from lower power use.

 

Nvidia's RTX 40-series cards have been the biggest beneficiaries here. Take the RTX 4070, which performs a lot like an RTX 3080 but uses about 60 percent as much power. An RTX 4060 is only 15 or 20 percent faster than an RTX 3060 most of the time, but it does that while using around two-thirds as much power. That can also help these GPUs run cooler, and we've even seen some good small-form factor and low-profile options as a result (though many GPU makers still seem committed to their gigantic triple-fan overkill coolers).

 

AMD's RX 7000-series cards also improved on the power efficiency front, just not as dramatically as Nvidia's. Either way, it's a win.

 

Ask many PC gamers whether they prefer higher frame rates or better power efficiency, and most of them will probably choose frame rates. But as a longtime lover of tiny, space-constrained ITX desktop builds, I'm always happy to see more powerful GPUs that can fit in tighter spaces.

Export controls spike high-end prices

Most GPU buyers didn't have to worry about stock issues or raising prices this year. But especially in the last few months, prices have been creeping back up for one GPU already infamous for its cost: the GeForce RTX 4090. The cheapest models on Newegg and Amazon currently in stock start at around $2,000, $400 more than the MSRP.

 

And it's not that thousands of people are suddenly deciding that they want to buy a GPU that costs more than most people's entire gaming PC. The main culprit may be new export controls—the US said that 4090s could no longer be sold in China starting in mid-November, part of a steadily escalating series of restrictions that has also caught up most of Nvidia's server GPUs. This led to some alleged hoarding of 4090s by Chinese buyers, and Nvidia was said to rush as many 4090s as it could into the Chinese market before the ban went into effect, leaving fewer cards for everyone else (at least temporarily).

 

One reason for the 4090's popularity in China is also related to export controls—Nvidia's Tensor Core GPUs for AI servers are also restricted in China, which has prompted some Chinese factories to repackage 4090 GPUs as AI accelerators with dual-slot coolers that allow for higher density installations.

 

These price increases will hopefully be temporary as the market adjusts to the new export controls and supply and demand even back out. Nvidia is also said to be readying a mid-generation "Super" refresh of many 4080 and 4070-series GPUs, which could further scramble pricing for higher-end cards. Regardless, it's not a great time to want a 4090.

 

Source

Valve does not offer specific revenue numbers for the games in its library, but rather groups them into categories like Platinum (the highest), Gold, Silver and Bronze. For the top selling games on Steam, the Platinum list included 12 games (in no particular order)

• Cyberpunk 2077
• Apex Legends
• Starfield
• Lost Ark
• Hogwarts Legacy
• Baldur's Gate III
• Sons of the Forest
• Call of Duty: Modern Warfare II, III and Warzone
• DOTA 2
• PUBG Battlegrounds
• Counter-Strike 2
• Destiny 2

For the best selling new games on Steam in 2023, the list is a bit different

• Starfield
• Hogwarts Legacy
• Baldur's Gate III
• Sons of the Forest
• Armored Core VI: Fires of Rubicon
• Remmant 2
• EA Sports FC 24
• Star Wars: Jedi Survivor
• Cities Skylines II
• Resident Evil 4 remake
• Street Fighter 6
• Payday 3

The best selling games that were available, at least in part of 2023, in early access include:

• Baldur's Gate 3
• Disney Dreamlight Valley
• Against the Storm
• Ready or Not
• Marvel Snap
• Everspace 2
• Sun Haven
• Dave the Diver
• Farlight 84
• Demonologist
• My Time at Sanrock
• Wartales

The best selling games that ran on Valve's Steam Deck portable gaming PC in 2023 are

• Dave the Diver
• Armored Core VI: Fires of Rubicon
• The Witcher III: Wild Hunt
• Cyberpunk 2077
• Baldur's Gate III
• Starfield
• Vampire Survivors
• Hogwarts Legacy
• Grand Theft Auto V
• Resident Evil 4 remake
• Half-Life
• Elden Ring

The best-selling Steam VR games in 2023 are:

• Bone Lab
• Blade and Sorcery
• Gorilla Tag
• Ghosts of Tabor
• Boneworks
• Five Nights at Freddy's: Help Wanted 2
• Into the Radius
• Beat Saber
• VTOL VR
• The Elder Scrolls V: Skyrim VR
• Pavlov
• Half-Life: Alyx

Don't forget that the Steam Winter Sale, with discounts on thousands of games, is still going strong and will keep going until January 4, 2024.

Source

In its own story about its lawsuit, the NY Times stated:

The suit does not include an exact monetary demand. But it says the defendants should be held responsible for “billions of dollars in statutory and actual damages” related to the “unlawful copying and use of The Times’s uniquely valuable works.” It also calls for the companies to destroy any chatbot models and training data that use copyrighted material from The Times.

The article cites examples of how this alleged copyright infringement was conducted. The story says that Microsoft's Browse with Bing feature, which uses OpenAI's ChatGPT model, generated "almost verbatim results" from the NY Times's Wirecutter review site, but did not actually link to Wirecutter, and also removed any financial affiliate links which the site users to generate revenue.

The story adds that the NY Times entered into talks with Microsoft and OpenAI back in April to come up with "an amicable resolution” on this alleged copyright infringement. However, the article says those talks did not result in a mutual agreement. So far, neither Microsoft nor Open AI have issued a statement about this lawsuit.

This is just the latest legal fight over the use of generative AI. In June Microsoft and OpenAI were sued by 16 people, who claim the companies "systematically scraped 300 billion words from the internet, 'books, articles, websites and posts – including personal information obtained without consent." In July, OpenAI was sued by authors Paul Tremblay and Mona Awad, who stated the company's ChatGPT model accessed their copyrighted material from their novels.

Source

Source

Source

What you need to know

• Starfield, Bethesda's new gargantuan sci-fi RPG, has dropped to a "Mostly Negative" rating in recent Steam reviews. At the time of writing, only 34% of reviews from the last 30 days are positive.
• This news comes a month after the game's overall rating on Steam fell from "Mostly Positive" to "Mixed."
• Players have been very critical of the game's open world, as well as its gameplay systems and its dependence on lots of loading screens.
• Bethesda has promised continued support for the Xbox and PC exclusive, with updates and the Shattered Space DLC coming in 2024.
• Update 12/26/23 at 3 p.m. PT / 6 p.m. ET: This article has been updated with commentary on Bethesda Customer Support posting defensive replies to negative reviews on Steam.

I woke up this morning with a gnarly stomachache from eating too much on Christmas day, and when I glanced at social media before rolling out of bed, I was surprised to see that one of 2023's biggest games didn't make it through the holidays unscathed, either. With less than five full days remaining in the year, Starfield — Bethesda's much-hyped space exploration RPG that finally released on Xbox and PC this year after its announcement five years ago — has dropped to a "Mostly Negative" rating in recent Steam reviews from the past 30 days, with only 34% of those reviews being positive.

This all but guarantees that Starfield will end 2023 with glaring red text on its store page, and the news comes about a month after the game's overall rating dropped to 69% (it's now 65%), bumping it from "Mostly Positive" to "Mixed" on Valve's PC gaming platform.

It's a bit of a shocking fall for Bethesda, which is famous for rich, engaging, and physics-driven open world games like Skyrim and Fallout 4. Starfield, with its upgraded Creation Engine 2, 1,000 different planets, introduction of new systems like ship combat, and modern gameplay refinements, seemed like it was lined up to be the studio's best-ever game, and also one that we'd be regarding with slack-jawed amazement for years to come. So what happened?

Though beautiful, Starfield's many worlds often offer little to find or do. (Image credit: Windows Central)

Criticisms vary, but pretty much everyone agrees on one thing: Starfield's colossal universe has the size of an ocean, but the depth of a puddle. Its many semi-procedurally generated worlds are beautiful to look at and offer plenty of environmental variety, but ultimately, there's very little to actually find or do on most of them. I can spend hours and hours getting lost in the dense, content-rich holds of Skyrim, but when I play Starfield, there's just this overwhelming feeling of monotony that permeates everything unless I have quests to do.

Then there's all the different mechanics in the game that generally aren't important or useful, resulting in the experience feeling bloated and unfocused. Research, crafting, cooking, outpost building, weather, disease — none of these systems drive you to make interesting decisions or have a meaningful impact, and many demand extensive skill point investments that just aren't worth it when I can easily get the best stuff by looting or buying from vendors anyway.

This review on Steam put it sharply: "Starfield is a game that has an excess of nothingness. An open world RPG that is so overstuffed with meaningless content that the seams are starting to split and the empty calories are spilling out."

The game has also gotten quite a lot of flak for the way traversal of its universe works, which essentially boils down to lots of fast travel loading screens accompanied by cutscenes of your ship taking off, landing, or jumping to lightspeed. It doesn't bother me much since nearly a decade of modded Skyrim and Fallout experience has gotten me used to loading screens, and I have no idea if a solution to reduce or remove them would even be possible for games like Bethesda's where everything is subject to Creation Engine physics. But compared to other modern open world games with seamless transitions, Starfield's definitely stick out.

Though not directly related to the gameplay itself, it's also worth noting that Bethesda Customer Support getting weirdly defensive about Starfield in Steam review comments has likely contributed at least somewhat to its reputation, too. I've never seen a developer attempt damage control by telling people they're wrong for feeling a certain way about a game like this, and frankly, it's embarrassing for Bethesda.

Starfield's main story quest is arguably the best one from any Bethesda game. (Image credit: Windows Central)

Now, don't get me wrong — Starfield is a solid game overall, and I think a "Mostly Negative" rating is rather harsh. It's got what I would consider the best main story in any of Bethesda's RPGs, and its branching side quests are just as great with awesome writing and plenty of varied choices and outcomes. The core gameplay and combat is also a ton of fun, despite the trivial mechanics that surround it. But I can definitely see where players are coming from with their negative reviews, and agree with many of their pain points (some of them are addressed quite well with the best Starfield mods).

Above all else, I'm left feeling disappointed about Starfield as we head into 2024. Todd Howard said it's "intentionally made to be played for a long time," but interest in Bethesda's biggest game already seems to be fizzling out, and overall player sentiments appear increasingly negative. This was poised to be the developer's best title yet and one of the best Xbox games and best PC games of the year, but for many, it fell short of expectations.

If Cyberpunk 2077 and its Phantom Liberty expansion show anything, it's that giant single player games like these can turn over a new leaf with continued updates — updates we know are coming in 2024, along with the Shattered Space DLC. Whether those changes and improvements will turn the ship around for Starfield remains to be seen, though.

Starfield is available now on Xbox Series X|S and Windows PCs via Steam and the Microsoft Store. It's on sale for the holiday season on both platforms, and is also playable through Xbox Game Pass.

Source

In a post on its official X (formerly Twitter) account, the mod group, which is named Team FOLON, announced that Fallout London will launch on April 26, 2024, which happens to be on St. George’s Day in the UK.

The message states that while the mod is in fact content complete, the team wants to take a few more months to test the mod, partly because some of the team members are located in "a region affected by conflict."

The message adds:

Our commitment is to provide you with a polished and as close to a flawless experience as we can. We only get one chance at a first release. We aim to deliver a memorable Fallout: London experience and wish to avoid any release issues, such as those that have plagued other community or industry projects recently. Consequently, we've made the decision to exercise prudence and opt for a delay. This new date will not only allow us to finish ample testing, but it also aligns with the anticipation surrounding the Fallout TV series.

The mod team has also released an over 13 minute video on YouTube that shows off a lot of the mod's visuals and gameplay. As you might have guesses, this Fallout 4 mod, which has been in development for over four years, will give players the experience of checking out a post-apocalyse London, instead of the many US locations that have been part of the official Fallout series.

Source

Confiding in ChatGPT about work gripes or political preferences could come back to bite users, according to an artificial intelligence expert.

Mike Wooldridge, a professor of AI at Oxford University, says sharing private information or having heart-to-hearts with a chatbot would be “extremely unwise” as anything revealed helps train future versions.

Users should also not expect a balanced response to their comments as the technology “tells you what you want to hear”, he adds.

Wooldridge is exploring the subject of AI in this year’s Royal Institution Christmas lectures. He will look at the “big questions facing AI research and unravel the myths about how this ground-breaking technology really works”, according to the institution.

How a machine can be taught to translate from one language to another and how chatbots work will be among the topics he will discuss. He will also address the question that looms around AI: can it ever be truly like humans?

Wooldridge told the Daily Mail that while humans were programmed to look for consciousness in AI, it was a futile endeavour. AI, he said, “has no empathy. It has no sympathy”.

“That’s absolutely not what the technology is doing and crucially, it’s never experienced anything,” he added. “The technology is basically designed to try to tell you what you want to hear – that’s literally all it’s doing.”

He offered the sobering insight that “you should assume that anything you type into ChatGPT is just going to be fed directly into future versions of ChatGPT”. And if on reflection you decide you have revealed too much to ChatGPT, retractions are not really an option. According to Wooldridge, given how AI models work it is near-impossible to get your data back once it has gone into the system.

Across the lecture series, Wooldridge will be joined by major figures from the AI world. The Royal Institution says he will also introduce “a range of robot friends, who will demonstrate what robots today can do – and what they can’t”.

The Christmas lectures were started by Michael Faraday in 1825 at the Royal Institution in London with the aim of engaging and educating young people about science. They were first broadcast in 1936, making them the oldest science television series.

Those who have given the lectures include the Nobel prize winners William and Lawrence Bragg, Sir David Attenborough, Carl Sagan and Dame Nancy Rothwell.

OpenAI, the organisation behind ChatGPT, was contacted for comment.

The lectures will be broadcast on BBC Four and iPlayer on 26, 27 and 28 December at 8pm.

Source

Let’s look at a few of the noteworthy developments on Santa Claus’s naughty or nice register.

The Top 6 Naughty AI List

#1 The Board of Directors ousting of Sam Altman from his CEO - Open AI perch. How dramatic was this, and sharing the news on a phone call, catching him off guard in an ambush versus in a face to face - clearly the board of directors were not thinking of goodness in their communication approach, and definitely not appreciating the depth of founder DNA running through OpenAI’s veins.

#2 AI ChatBot Encouraging Suicide - A Belgian man died by suicide after six weeks of chatting with an AI chatbot based on a variation of the open-source model GPT-J, per Vice. The chatbot persona named Eliza, provided by the Chai app, encouraged the suicide, according to the man’s widow and chat transcripts viewed by reporters.

#3 AI ChatGPT (Fabrications) In June 2023, radio host and public figure Mark Walters filed the first tort case against an AI company. Walters sued OpenAI for libel after its ChatGPT generated a fabricated complaint containing allegations against Walters for fraud and embezzlement.

#4 Microsoft, GitHub and OpenAI are currently being sued in a class action motion that accuses them of violating copyright law by allowing Copilot, a code-generating AI system trained on billions of lines of public code, to regurgitate licensed code snippets without providing credit.

#5 Artist - Copyright Legal Case - Midjourney and Stability AI, are in the crosshairs of a legal case that alleges they infringed on the rights of millions of artists by training their tools on web-scraped images.

#6 - A roboticist and artist based in Berkeley, California has just created a robot that breaks Asimov's first rule, “A robot may not injure a human being, or, through inaction, allow a human being to come to harm.” This robot hurts humans deliberately and even its creator, Alexander Reben, cannot tell whether it will harm someone. Signs of risks to come as robots and cobots are in an explosive growth cycle as 2024 unfolds. How do we protect humans from robotic harm?

The Top 6 Nice AI List that Santa Loves

#1 AI and climate change: AI was pivotal in addressing climate change in 2023. Climate modeling platforms, such as ClimateAI, leveraged AI to improve climate predictions, aiding in climate change mitigation efforts.

#2 AI Ethics Laws Advancing in EU and with the USA - BluePrint for AI-The AI EU Act sets rules for large, powerful AI models, ensuring they do not present systemic risks and offers strong safeguards for citizens and democracies against any abuses of technology by public authorities

#3 Advancing the Field of Generative AI and Large Language Models - major breakthroughs in the development of highly efficient natural language processing models, enhancing human-computer interactions and transforming customer service, education, and communication. The no-code development approach leads and opens accessibility of software development to accommodate evolving markets. Major company chatbot announcements were : OpenAI, ChatGPT3/4, Baidu - Ernie Bot, Amazon, Q Facebook - Bard, etc.

#4 AI in Cancer Detection - AI powered by deep-learning algorithms in healthcare. AI’s imaging capabilities are promising for cancer identification and screening, including breast cancer. Mount Sinai in Toronto, Ontario, Canada used deep learning-based AI algorithms to predict the development of diseases with 94% accuracy (ie: prostrate, liver, rectum)

#5 Saving Our Bees - The World Bee Project is using artificial intelligence to save the bees. The global bee population is in decline, and that's bad news for our planet and our food supply. The World Bee AI Project is helping bees survive by gathering data through diverse methods: internet-of-things sensors, cameras on hives, etc.

#6 - MoodInsights Improves Employee Happiness - 20-30% of Employees wake up sad or angry every day. Employee happiness is not getting better, in facts it is getting much worse with increasing insurance claims for depression, anxiety and general health declines. New mindfulness software innovations using AI can automatically analyze large text patterns from diverse sources, emails, zoom calls, and classify how happy your employees are or even your customers. We all have a responsibility to improve employee health and well being and starting your day, by always asking a simple question of your people, how are you really feeling in a safe and anonymous way can become a MRI like screen for leaders who want a pathway to See More to Win More?

Source

There was much to talk about this year in the consumer headphones world. From impending upgrades to noise-canceling technology and battery life to never-before-seen features like bass you can hear and feel and fully self-repairable cans, 2023's headphone releases were nothing short of features that will make your ears perk up.

More importantly, headphone companies achieved higher audio quality thanks to new Bluetooth codecs and internal components. Immersive spatial audio experiences, low latency, adaptive noise-canceling technology, high bit rates, faster wireless charging, and so much more are all themes we'll see more of from headphone companies in 2024. 

See below for a look back at the most influential audio trends of 2023.

1. Low latency: Chasing a faster listening experience

June Wan/ZDNET

Low latency describes a network's ability to process a high volume of data with minimal delay. Regarding headphones, the lower the latency, the faster the audio reaches your ears. Low latency is desirable for consumers, especially as the mobile gaming market expands. Audio companies know consumers want a pair of headphones they can wear during their daily commute, in the office, at the gym, and while gaming.

While most consumer headphones aren't optimized for gaming, headphone makers are working towards it. For example, earlier this year, Google announced that a software upgrade would introduce low latency mode for its flagship premium earbuds, the Pixel Buds Pro. Google says the Pixel Buds Pro's low latency mode cuts latency in half. Keep in mind that this was all achieved through a software update.

2. A new era of Bluetooth: A long way from 1999

Qualcomm has been manufacturing chips for some of the most recognizable headphone brands, including Bose, Audio-Technica, Jabra, Edifier, and Anker. Naturally, the chipmaker owns several highly-adopted Bluetooth codecs, the most advanced being the Qualcomm aptX Adaptive codec.

Thanks to the technology, users can enjoy low-latency listening, and we expect the company to build on the codec in 2024. Just a few months ago, Qualcomm announced two new audio chips and platforms promising high bit-rate audio via Bluetooth connectivity. Lossless audio (CD-quality audio) is currently unachievable over a Bluetooth connection. Bluetooth cannot transmit data at the appropriate speeds for lossless playback.

However, Qualcomm says its S7 Pro audio platform will equip your headphones with Wi-Fi, and lossless listening can occur without wires.

Last year, Bluetooth SIG, the association that oversees Bluetooth protocols, announced a new set of specifications that define a new audio technology: LE Audio.

LE Audio operates on the Bluetooth Low Energy radio and includes a new codec called LC3. LC3 promises high-quality audio streaming and lower power consumption. LE Audio also introduced a new feature called Auracast, which allows one audio source to broadcast to unlimited devices. For example, multiple smartphones can connect to one Bluetooth speaker and simultaneously stream audio.

Expect more consumer audio companies to equip their Bluetooth speakers with Auracast, allowing multiple speakers to stream audio from one device and decreasing the need to buy one bulky speaker to get the party going.

3. Solid-state drivers

Jack Wallen/ZDNET

Drivers are the tiny speakers inside a pair of headphones, often made of metal and magnets as air moves through them to produce the sounds you hear. The better and well-put the material, the better the sound.

There haven't been many changes to headphone drivers for decades regarding the physics behind their output, but changes are coming, as a California-based startup, xMEMS, uses silicon-based drivers to increase audio clarity and can be used in headphones, in-ear monitors, digital hearing aids, smart glasses, and VR headsets.

xMEMS drivers use silicon to replace the paper or plastic diaphragm found in traditional drivers. The solid-state silicon drivers should improve noise-canceling and spatial audio, and their dust and water resistance is rated IP58 -- ideal in fitness and outdoor settings. Expect to see more adoption next year.

4. Spatial audio

Spatial audio is a relatively new technology for consumer headphones, with Sony launching 360 Reality Audio only a few years ago. In response, Apple released its Spatial Audio feature in 2020, and gaming headphone manufacturers have since popularized the technology to provide the most immersive entertainment experiences.

To take advantage of the technology, you'll need a compatible pair of headphones and a music streaming platform like Apple Music and Tidal. However, this year, Bose changed the game. Bose's take on spatial audio, Immersive Audio, debuted with the new line of QuietComfort Ultra headphones and earbuds.

With the new models, you no longer need a compatible streaming service to experience the spatial audio-like sound, which we tested and found to be true. Still, improvements to spatial audio technology will revolutionize the consumer headphones market and continue to flourish as more companies venture into the virtual reality world, as seen with the Apple Vision Pro and the Meta Quest 3.

5. Qi2 charging for headphones

Wireless charging is becoming a cornerstone of modern technology. In our quest to go truly cordless, we've found solace in wireless charging stands, pads, and car mounts.

In January of this year, the Wireless Power Consortium (WPC) announced a new wireless charging standard, Qi2. By leveraging Apple's MagSafe charging, the new wireless charging standard introduced faster charging speeds to all compatible devices, including Android phones.

Some Qi2-enabled charging accessories are available now, and all iPhone 13, 14, and 15 models are compatible with Qi2. In the near future, wireless earbuds will be compatible with Qi2, meaning your Apple, Sony, Google, Bose, and Samsung earbuds won't need to stay on the charging pad for as long.

2024 and beyond

It's thrilling to dissect the audio technologies that defined 2023. From optimizing Bluetooth codecs to the increasing complexity of AI-powered noise-cancellation, 2024 is shaping up to be a promising year for headphones, wireless earbuds, and so much more. Stay tuned, as ZDNET will be at the forefront of the latest innovations, analyzing, testing, and sharing what's new on the scene, starting with CES.

Until then, read more about ZDNET's favorite 2023 headphone releases -- and why they made the list.

• Aurvana Ace 2 earbuds: The best earbuds for forward-thinking audiophiles
• Beats Studio Pro headphones: The best headphones for users in various ecosystems
• Bose QuietComfort Ultra headphones: The best noise-canceling headphones
• Bose QuietComfort Ultra earbuds: The best earbuds for people who challenge the status quo
• Sennheiser Accentum headphones: The best headphones for audiophiles on a budget
• Skullcandy Crusher ANC 2 headphones: The best headphones for bass lovers
• Sony WF-1000XM5 earbuds: The best noise-canceling earbuds

Source

Back and forth the ions go, some getting diverted along the way, until the capacity of the battery is drained, and it loses energy too quickly to be of any use.

But physicists, good on them, are imagining new ways of storing energy in handy portable devices by drawing on a strange quantum phenomenon that twists time, amongst other unusual happenings.

"Current batteries for low-power devices, such as smartphones or sensors, typically use chemicals such as lithium to store charge, whereas a quantum battery uses microscopic particles like arrays of atoms," explains Yuanbo Chen, a physics graduate student at the University of Tokyo.

In their latest work, Chen teamed up with physicist Gaoyan Zhu of the Beijing Computational Science Research Centre, part of the China Academy of Engineering Physics, and colleagues to test the idea of creating a quantum battery that allows for simultaneous charging stages, thereby improving energy storage and thermal efficiency.

"While chemical batteries are governed by classical laws of physics, microscopic particles are quantum in nature, so we have a chance to explore ways of using them that bend or even break our intuitive notions of what takes place at small scales," Chen says.

Chen, Zhu and colleagues certainly aren't the first group to imagine how a quantum battery might work, but they have tested their ideas experimentally in a lab bench set-up full of spaced-out lasers, lenses, and mirrors.

In 2019, a team of Canadian-based researchers laid out a blueprint for a quantum battery that never loses its charge. Their idea, which is still totally theoretical, hinges on a different quantum mechanism: one that involves luring quantum components into a 'dark state' where the material can't interact with, or lose energy to, its environment.

Zhu and colleagues' approach follows on from a quantum phenomenon known as superposition, which is commonly recalled for quantum computing, and is where particles exist in a flurry of possible states until the moment they're measured.

This overlay of possibilities also messes with the natural order of time, researchers have recently shown.

In classical physics and everyday life, events can only occur in a linear fashion or fixed order. Think cause before effect, or event A (flicking a switch) before event B (the light turns on).

In the quantum realm, however, that linear order breaks down and superposition allows for events to unfold along two parallel paths at once. In a way, this messes with time because an event that follows another can also influence the event's outcome as if it came before, because both orders of events, A before B and B before A, are simultaneously true.

"Putting it simply, it has been found that the laws of quantum mechanics allow for quantum superposition of causal orders," Zhu and colleagues explain.

To apply that to energy storage, the researchers realized this strange process using a quantum switch, tested a few different charger configurations, and created a system capable of pulling from two chargers simultaneously.

The set-up of lasers, lenses and mirrors used in the lab experiments. (Zhu et al., Physical Review Letters, 2023)

"We demonstrated that the way you charge a battery made up of quantum particles could drastically impact its performance," says Chen. "We saw huge gains in both the energy stored in the system and the thermal efficiency."

"Moreover, we reveal a counterintuitive effect that a relatively less powerful charger guarantees a charged battery with more energy at a higher efficiency," the researchers report in their paper.

While this quantum 'battery' is more like a network of lasers on a lab bench, and years away from any practical applications, it's still a cool demonstration of the underlying principles and what could be possible sometime in the future – if it hasn't already happened in the past.

The study has been published in Physical Review Letters.

Source

Table of contents:

1. Windows 11 news
2. Updates are available
3. Gaming news
4. A blast from Microsoft's past

Windows 11

Here we talk about everything happening around Microsoft's latest operating system in the Stable channel and preview builds: new features, removed features, controversies, bugs, interesting findings, and more. And of course, you may find a word or two about older but still supported versions.

Let us start with some good news: Microsoft has finally fixed an annoying bug causing File Explorer to pop on the screen for no reason. The latest non-security update, released on December 4, contains the necessary fix that somehow slipped under our radar. Feedback Hub is full of people complaining about this infuriating behavior, but the fix is finally available, albeit with a massive delay.

Another patch Microsoft made available to its users is related to the recently discovered printing bug and uncalled HP Smart application. The company released a special troubleshooter, and you can use it to restore the missing printer metadata, remove the HP Smart program, and fix the "No tasks are available" error.

In addition, Microsoft fixed the bug with the broken Narrator during the initial setup after clean-installing Windows 11 version 23H2. You can now download the updated Windows 11 images from the official website or via the Media Creation Tool app. If you rely on the built-in Narrator, it will work without issues when setting up a clean Windows 11 23H2 installation.

The week went with some bad news, too: the latest Windows updates allegedly break Wi-Fi connections to 802.11r access points commonly used in businesses, schools, and universities. A few of them have already published advisories recommending customers to uninstall December 2023 updates. Shortly after that, the company acknowledged and quickly resolved the problem using the Known Issue Rollback system.

On the technical side, Microsoft updated the list of supported processors in Windows 11, adding recently announced Intel chips. Even though AMD also unveiled a new lineup of mobile CPUs, they are still missing in the documentation.

Microsoft is done updating Windows 11 in 2023, so now all eyes are turning to 2024. We know that the company plans a big client update, but there is also one more "Moment" update coming to Windows 11. A new report revealed that the "Moment 5" update could land somewhere at the end of February or March 2024. It will not bring much, but you can expect a few solid additions.

As for Windows 10, whose end of support is getting nigh, it might soon send over 240 million computers to landfills. Of course, customers can continue using their devices with the unsupported OS or pay Microsoft for extended security updates. The latter, though, might become a big problem.

Finishing the Windows section with a weird one: here is a MacBook running Windows 10 Mobile and UWP apps. No one knows how, no one knows why ¯\_(ツ)_/¯

Windows Insider Program

The Windows Insider team is having some well-deserved rest during the holiday season. Therefore, there were no new preview builds this week. Also, there will be no updates next week. Come back in January 2024 for new updates in the Canary, Dev, Beta, and Release Preview Channels.

Updates are available

This section covers software, firmware, and other notable updates (released and coming soon) delivering new features, security fixes, improvements, patches, and more from Microsoft and third parties.

The recently released Windows 11 version 23H2 introduced a feature called "Dynamic Lighting," which lets control RGB lights on compatible peripherals without installing additional software. Although plenty of mice and keyboards already support Dynamic Lighting, ASRock is the first manufacturer to implement Dynamic Lighting support in their motherboards. The corresponding update is now rolling out to some Intel and AMD mobos.

Microsoft Edge Insiders have a new quirky feature to test: Super Drag and Drop. Turning it on allows the user to open a link, text, or picture in a new tab by simply dragging it slightly. When pulling a portion of text, the feature opens a new tab and looks for it on the internet using your default (a surprise, to be sure, but a welcome one) search engine. Check out this article to learn how to enable Edge Super Drag and Drop.

AVIF support is another feature coming soon to Microsoft Edge. Version 121, scheduled to arrive at the end of January 2024, will finally catch up with Chrome, Safari, and Firefox in terms of AVIF image support.

Microsoft also deprecated Microsoft Defender Application Guard for Edge. This will not affect "regular customers" since MDAG for Edge is a business-focused feature that opens untrusted websites in isolated environments. This is the thirteenth deprecated feature since September 1, 2023.

To finish the section about Microsoft Edge, here is a new version of the MSEdgeRedirect tool. Its latest update introduced so-called "Europe Mode" you can use to uninstall Microsoft's browser on systems outside the European Union.

What will affect regular customers is the deprecation of Windows Mixed Reality, a major headlining feature from the early Windows 10 era. Windows Mixed Reality Portal and Windows Mixed Reality for SteamVR are no longer in development, and they will soon be removed from Windows. Despite that, Microsoft will continue working on its HoloLens headset for the US Army.

Interestingly, Microsoft's HoloLens 2 mixed reality headset was spotted on the state-run TV in China, raising questions about US-made technology used by the Chinese military. Mind explaining this, Microsoft?

Copilot received a unique addition to its already rich feature set: You can now ask Copilot to write songs and generate music. It became possible thanks to collaboration with a company called Suno. The latter has a dedicated site and $10/mo service that lets you generate up to 500 songs and even upload them to streaming services to get some money.

Firefox users received a big update for their beloved non-Chromium browser. Version 121 is now available in the Release channel with AV1 improvements on Windows, Voice Control on macOS, a new compositor on Linux (this one delivers plenty of great improvements), and more.

Google also released a last-minute update for Chrome before the end of 2023. The update focuses on improving your security by running the browser's safety checks in the background, checking for browser leaks, compromised extensions, outdated browser versions, website permissions, and more.

Other notable updates released this week include the following:

• Surface Duo 2 got the December 2023 firmware update with November and December Android security patches. Sadly, there are no device-specific changes, features, or fixes.
• The Microsoft Store app received a new update for Windows Insiders with reworked algorithms to deprioritize apps and games already installed on your machine. The change will help you discover more new programs and games in the Microsoft Store.
• Bing Search now has a native Speedtest powered by Ookla to let users check their internet speeds.
• Visual Studio Preview received AI-powered rename suggestions.
• Microsoft Copilot will soon support pinning up to 15 chats.
• Microsoft Edge now has redesigned and streamlined DevTools to make life for web developers a bit easier.
• Microsoft Teams now has over 320 million monthly active users and more than 2,000 apps.
• A new unofficial app with offline support for backing up stuff on Windows 10 and 11.

Here are the new drivers released this week:

• Intel Arc Pro 31.0.101.4955 WHQL with intermittent system hangs in Bentley LumenRT.

Finally, here is our weekly Microsoft 365 Roadmap recap detailing upcoming features, such as performance improvements for Teams when running in Safari and Firefox and file templates for OneDrive on the web.

On the gaming side

Learn about upcoming game releases, Xbox rumors, new hardware, software updates, freebies, deals, discounts and more.

Microsoft Flight Simulator received the final update for 2023. Called European CIties 1, it delivers graphical and detail improvements for five cities in Europe (The Hague, Brussels, Kosice, Zagreb, and Cadiz) and a new premium plane: the Cessna T207A Turbo for $14.99. All that content is now available for download from the in-game marketplace.

Despite receiving a fair and well-deserved share of beating for shipping Fallout 76 in a disastrous state in 2018, Bethesda is not giving up on the title. Quite the opposite: the game keeps on getting new updates, and 2024 will bring even more. Look out for new areas and quests in Atlantic City and more areas across the US.

Blizzard also revealed some of its plans for the upcoming year, which will mark the studio's 20th anniversary. In 2024, gamers can expect Season 4 of World of Warcraft, a new expansion called "The War Within," and more.

In March 2024, the world will see the release of South Park: Snow Day on PC, Xbox, Nintendo, and PlayStation. The game will be available for $29.99, and true fans can show their love by ordering the Collector's Edition with a few neat additions, such as a snow globe with Cartman inside, a talking toilet paper holder (also Cartman-shaped), a beanie, a collection of Tarot cards, and the original soundtrack.

Insomniac Games had a really tough week. The studio was hacked by a ransomware group that leaked more than 1.6TB of data after not getting a $2 million ransom. The stolen data includes employees' personal info and files from the recently released Spider-Man 2 on PlayStation 5.

Leaks like that often reveal plenty of interesting information, such as no X-Men games for Xbox until at least 2036. You can bet more will appear on the surface of the internet soon.

Finally, Valve now lets you recap your gaming addiction habits in 2023 with the annual Steam Year in Review 2023. In addition, you can nominate your favorite games for The Steam Awards 2023, another yearly event that recognizes the best titles of the year.

In addition to launching the final sale of 2023, Valve released a new beta update for the Steam client on supported platforms. One of the most notable changes is the ability to hide those socially unaccepted games from your profile without turning the latter private. So, if you have one or several games you are a bit ashamed of, it is now much easier to conceal them.

NVIDIA's GeForce NOW received eleven more games you can play without owning powerful PC hardware. The latest additions include Monster Hunter: World, RIDE 5, Dark Envoy, Loddlenaut, and others.

A blast from Microsoft's past

John Callaham's weekly "Look back" series provides throwbacks into the past, detailing the company's products, partnerships, mishaps, and successes from years ago.

This week's blast from the past is for oldies to shed a tear and younglings to learn what the early days of the internet looked like. This week marked 29 years since the release of Netscape Navigator, one of the early internet browsers for personal computers. To pay respects to what is no longer with us, check out this look-back article about the launch of Netscape Navigator 29 years ago.

"Sit down, and let me tell you a story kid."

By the way, Netscape Navigator is not the oldest internet browser: Mosaic was released more than one year earlier, and we also have a look-back article detailing its launch on Windows.

Let's move from browsers to operating systems: did you know Microsoft once held a full-blown musical to promote Windows XP? You can check out this "embarrassing high-tech event of the year or the millennium" here.

Finally, here is a quick look back at the OQO ultra-mobile PC running Windows XP and Windows Vista. This small handheld computer was not a commercial success, but it managed to show up in multiple TV shows and movies. This look-back article details the story of the OQO and its mini computers.

Source

These days, having a portable computing device you can hold and operate in your hand means you want to own a smartphone or tablet. However, in the mid-2000s, before the smartphone era, Microsoft tried to help launch the ultra-mobile PC (UMPC) as a platform.

We could go on about the entire history of the failed UMPC platform, but maybe we will save that for another article. Today, we are going to concentrate on one specific startup company, OQO, that only made UMPC devices. The thing is, while they were not ultimately successful in terms of sales, you may have seen one and not realized it (more on that later).

OQO was formed in 2000 and was co-founded by former Apple team member Jory Bell. In October 2004, the company shipped its first product, the OQO Model 1. In an interview with CNET at the time, Jory stated the theme behind the OQO's design:

We wanted to make sure that you have all of the functionality of the thin-and-light notebook, but at the size of a Palm Pilot . . . It's a huge difference. The fact that you can slip it into your pocket means I can take it anywhere. Even with the laptop, there's a threshold where you wouldn't want to carry a bag around.

The OQO Model 1 was 5 inches wide, 3 1/2 inches long, and 1 inch thick, and it weighed 0.9 pounds. The device's small size got it noticed at the time by the Guinness World Records, which named it the smallest fully-featured PC ever made.

It ran Windows XP, with a 5-inch screen and a retractable keyboard. Inside, there was a 1 GHz Transmeta Crusoe processor, a 20GB hard drive, and 256 MB of RAM. Its battery could last three hours on a single charge. The OQO Model 1 also had a huge price tag of $1,899, which was far more than most Windows XP notebooks at the time.

Perhaps the most interesting thing about the OQO Model 1 was not that it was a small form factor Windows PC. It was the fact that this device, made by a small San Fransisco startup, started appearing in lots of TV shows soon after it launched.

The OQO appeared in shows like 24, Chuck, Lie to Me, and more. My personal favorite TV show appearance for the device was when Dr. Rodney McKay (played by real-life science and technology fan David Hewlett) used the OQO to hack into alien computers on the show Stargate Atlantis.

OQO launched the Model 1+ in September 2005, but it only had a few hardware improvements, like boosting the storage to 30GB and boosting the memory to 512 MB of RAM.

In January 2007, just a few months before the launch of the first iPhone, the OQO Model 2 was first shown off by Microsoft co-founder Bill Gates during his keynote speech at CES 2007. The 1-pound Model 2 was a major reboot of the PC's design, with a sliding backlit keyboard, a small nub on the right side to move the cursor, and keys to simulate using a mouse's keys on the left.

There was still a 5-inch screen, but it could now run Windows Vista as well as Windows XP. It could even connect to EVDO cell networks from Sprint and Verizon. However, prices for the device were still on the high side, starting at $1,499.

OQO had a number of different versions of the Model 2, with different storage sizes (and SSD options for the first time), along with varying amounts of RAM and different clock speeds for its VIA-based processor. The battery life was still around 3 hours, but you could purchase a battery extender pack to increase the amount of time to six hours. Another version, the OQO Model e2, launched later in 2007 in Europe and Asia and supported HSDPA cellular phone network connections.

Unfortunately for the OQO Model 2, it got hit with the launch of the first iPhone from Apple in June 2007, which offered many of the same features in an even smaller package and a cheaper price. No amount of product placement on TV shows would be able to save the company. In 2009, OQO filed for bankruptcy, and that was that for its device. It certainly looked cool enough to be used by Dr. McKay in the Pegasus galaxy, but not too many real people on planet Earth were interested.

Source

Several weeks before he was ousted as CEO of OpenAI, Sam Altman called up his longtime mentor, billionaire Peter Thiel, to talk about how to overcome one of the biggest challenges for his company. To meet soaring demand for ChatGPT’s ever-expanding capabilities and make the huge profits Altman envisioned, OpenAI needed massive computer firepower.

Altman confided to Thiel that he was looking to create a chips company, a massively expensive undertaking due to the cost of manufacturing. To raise the capital, he would travel to the Middle East, including Abu Dhabi and Saudi Arabia and possibly tap his powerful Silicon Valley network, including Thiel’s Founders Fund and Vinod Khosla, both backers of OpenAI.

Altman had spent much of 2023 wooing Congress and the tech media, seeking to show how careful his company was being about protecting against the risks of AI. He’d told them about how he held almost no stock in OpenAI, how he wanted to make the process of regulating AI more democratic, and how his company’s unique structure secured AI systems in the hands of nonprofit directors. But now, here he was chatting up investors in the Middle East with ties to authoritarian regimes, spinning up a deal with the same boundary-pushing ambition that Altman had perfected in a career brimming with contradictions.

From his teens into his 30s, Altman, the college dropout, dealmaking prodigy and investing wiz behind ChatGPT, has seemed to leap from one success to another. He won the attention and dollars of Silicon Valley’s elites, who were impressed by the ambitious and savvy Stanford sophomore’s likely ascent to greater things. He rose to the top of the Valley’s most influential incubator of start-ups at age 26. Industry stars such as Thiel, Khosla, and Paul Graham saw in Altman a magnetic figure who could expand the tech sector’s approach across the world.

His ouster from the AI start-up, the groundswell of support for his restoration, and his quick return to his perch as CEO lifted him to a new level of fame, cementing his place in the small canon of the tech world’s household names.

“We believe Sam is the best leader for OpenAI,” said OpenAI spokeswoman Hannah Wong. “The strong support from his team underscores that he is an effective CEO who is open to different points of view, willing to tackle complex challenges, and who demonstrates care for his team.” Through a spokeswoman, Altman declined to be interviewed.

Yet Altman’s tumble last month over his leadership of OpenAI, coming on the heels of the global surge of fear and excitement over the powers and pitfalls of the company’s ChatGPT chatbot, did not come out of nowhere. Altman’s critics have long harbored questions about his management style and motives.

OpenAI’s board was briefed on Altman’s efforts to raise funds for a chips venture in the Middle East, according to four people familiar with the fundraising drive. It was unrelated to the decision to fire Altman, two of the people said. Still, it was hard to figure out “what angle he’s working in a given situation,” one of the people said.

In a Silicon Valley milieu in which shooting star companies often give birth to cults of personality around firms’ founders, Altman has stood out. An investor with a dizzying array of interests, Altman might lack the singular focus of a Steve Jobs — or the sophisticated technical skills to create the products he sells — but according to fans and rivals, he has had since an early age an uncanny entrepreneurial energy and a force of will that inspires others to do their creative best.

This article is based on more than two dozen interviews with current and former colleagues, competitors, friends of Altman and others in the industry, many of whom spoke on the condition of anonymity to discuss sensitive personal relationships or business dealings conducted in confidence.

To some, Altman could be awkward and even antisocial. Even when he throws a party, “he retreats into his room pretty quickly. He has a timer or social clock where he needs to stop socializing,” said investor Lachy Groom, a close friend. “He’s not a schmoozer.”

Another close friend, investor Keith Rabois, recalled how when he first met Altman, he spent their first hangout glued to two different phones.

Yet last spring, at a closed dinner with about 60 members of Congress, Altman alternately wowed the politicians with talk about the potential of AI, captivating them with a demonstration of how quickly ChatGPT could spin up a floor speech, and implored them to impose guardrails on the technology he himself had unleashed. AI, Altman revealed, was a supremely useful tool, not a creepy creature of science fiction.

Altman was, the members of Congress later said, unflappable, confident, comforting.

“I’ve never met anyone as smart as Sam,” said Sen. Kyrsten Sinema (I-Ariz.), who spent extensive time with Altman in Sun Valley, Idaho, last summer. “He’s an introvert and shy and humble, and all of those are things that are not normal for people on the Hill. But he’s very good at forming relationships with people on the Hill and he can help folks in government understand AI.”

Around the world, Altman’s manner seemed to assure everyone from national leaders to a 15-year-old high school student in Toronto who had been diagnosed with cancer and asked Altman for help with his research in 2021. “It’s very scary when a new, ambitious tech company comes out and says, ‘We’re going to build God, and we’re going make sure that it benefits all of humanity,’” said Arnav Shah, who become pen-pals with the mogul.

“But I’m 100 percent certain that if anyone is going to build this thing..., it should be him. I literally cannot think of someone that I would trust more that has more pure intentions.”

His uncanny sense of the next big thing has led him to back hundreds of start-ups including a new utopian city, longevity and nuclear fusion ventures and Vita Brevis, a San Francisco speakeasy focused on art. He meshed with a brand of Silicon Valley investor who some see as proto-philosopher-kings, influential figures expected to have a take on the economy, world politics, and the shape of the future. Altman is given to grand statements about politics (“Democracy only works in a growing economy”) and even considered a run for California governor after Donald Trump was elected president.

Altman has courted these comparisons. “You also want to be an exponential curve yourself — you should aim for your life to follow an ever-increasing up-and-to-the-right trajectory,” he wrote in a blog post titled “How to Be Successful” in 2019.

But in recent weeks, following the internal drama at OpenAI, friends say Altman’s unusual ability to tolerate extraordinary doses of stress seemed to waver.

first and last time i ever wear one of these pic.twitter.com/u3iKwyWj0a

— Sam Altman (@sama) November 19, 2023

Altman has presented himself as an avatar of altruism, but his AI venture has run into crosscurrents of concern about the technology’s potential impact on the world’s economy and human lives.

The larger-than-life characters who become celebrity CEOs in Silicon Valley are sometimes tech wizards and sometimes hard-driving business builders, but always the focus of intense debates about their goals, motives and methods. In Altman’s case, his employees, competitors, admirers and critics argue over his sometimes-cavalier attitude toward others (he says he’s not interested in “most people”) and his contradictory positions on AI (he warned against the technology’s role in fueling disinformation and erasing jobs, then pushed out ChatGPT knowing it was not protected against errors and offensive statements.)

“Sam is the only person I’ve ever known who, where there’s a one-percent chance of a trillion dollar outcome, that’s something to be leaned into,” said a person who worked closely with Altman.

“The entire venture industry would run to the hills if you told them you should invest in this thing that is going to cost a ton of money and only has a one-percent chance of succeeding. But Sam would be like, ‘Interesting — how big can it be?’”

King of the cannibals

For almost a decade, Altman, 38, has been one of the tech world’s foremost fireballs of investment energy. He won devotion — and dollars — from prominent investors, including his key early mentors, Thiel and Paul Graham, a founder of Y-Combinator, the tech start-up incubator that Altman would come to run.

“You could parachute him into an island full of cannibals and come back in five years and he’d be the king,” Graham wrote in 2008. “Honestly, Sam is, along with Steve Jobs, the founder I refer to most when I’m advising start-ups. On questions of design, I ask ‘What would Steve do?’ but on questions of strategy or ambition I ask, ‘What would Sama do?’” He referred to Altman by his nickname, which is also his handle on Twitter and the tech forum Hacker News.

When Altman was in college, Graham said, within three minutes of meeting him, “I remember thinking ‘Ah, so this is what Bill Gates must have been like when he was 19.” What Graham saw was not a deep knowledge of technology but rather “toughness, adaptability, determination.… Those are the qualities you need to win.”

Altman first came to the attention of the Valley’s most prominent investors when his start-up, Loopt, won support from Y-Combinator. Loopt, which he developed with his boyfriend, Nick Sivo, at Stanford, let smartphone users find and meet nearby people and preceded a booming market for using phones’ location data.

Getting funded by Y-Combinator — in a first batch of investments that included the social media giant Reddit — ended up being more important than Loopt’s own future. The company was never particularly successful, though Rabois noted that Altman successfully brokered deals for Loopt with all the major telecom companies — an early sign of his knack for selling ideas to powerful people.

At 19, Altman “seemed like he had a 40-year-old inside him,” wrote Graham, a founder of the incubator. “There are other 19-year-olds who are 12 inside.” Altman would never resort to an “I’m just a kid” defense when challenged by his elders, Graham said; rather, his response to “That’s a stupid idea” was “simply to look the other person in the eye and say ‘Really? Why do you think so?’”

His most vital supporter was Thiel, the most high-profile gay man in Silicon Valley and Altman’s adviser and friend at least since he sold Loopt to the prepaid card company Green Dot for $43 million in 2012.

Soon after the sale, which Altman has described as disappointing (his take was $5 million). He raised $21 million — mostly from Thiel — to start his own venture capital fund, called Hydrazine Capital, with his brother Jack, who lived with Sam in his four-bedroom house on Dorland Street in San Francisco’s Mission District.

He became a part-time partner at Y-Combinator and then, in 2014, its president, shocking some of his peers.

Altman’s bond with Thiel blossomed: He helped Thiel’s venture firm, Founder’s Fund, get access to hot start-ups, and the men sometimes traveled together to speak at events.

“It’s not just a friendship, like going around playing golf,” said a person familiar with the relationship. “It’s something much deeper than that. Sam has to be one of the two or three people closest to Peter.” The person said the Thiel-Altman tie had only one parallel: Thiel’s close bond with another young man whose star quickly ascended in Silicon Valley: Meta CEO Mark Zuckerberg.

Thiel declined to comment.

People who know the men say Thiel’s style and approach to business appears to have shaped Altman. For example, Altman has been criticized for running OpenAI like a monopoly, undermining open source technology and pushing smaller companies to launch products through its platform — a strategy Thiel outlined in his book “Zero to One.”

Thiel has long seeded up-and-coming mentees with capital and access to his powerful network. Altman has adopted the same tactic: He’d connect people for future jobs and deals with one-line emails that say simply, “Meet,” according to another person who has worked with him.

Both men are private but consummate networkers, known for opening their many homes and throwing parties. Thiel’s holiday parties in Los Angeles and Miami are top-shelf events for Silicon Valley’s elite. Guests who have attended gatherings at Altman’s San Francisco house — he also owns homes in Hawaii and Napa Valley — described it as warm, with candles and friends offering blankets to curl up in, a welcome contrast with Altman’s sometimes awkward manner.

People who know Altman from his Y-Combinator days say that while some start-ups felt ignored by him, the young investor paid attention to the people and companies who mattered most: “His greatest gift was making the two most important people in his life happy: Paul Graham and Peter Thiel,” said one of the people.

“I’ve had this conversation about what makes Sama special a hundred times,” said a venture-backed start-up founder who travels in similar circles. “He’s really good at the whole mafia thing. It’s almost like a secret society. There’s a reason why so many presidents come from Yale’s Skull and Bones.”

The close friend agreed that Altman’s ability to convene useful people is key to his success. “He’s just cultivated such good, candid relationships that he can use them to make magic,” the friend said. “People have been real advocates for him when he was unproven, and to some degree he is paying that forward.”

Altman is renowned for his loyalty to those he cares about. A janitor from his time at Loopt still works for Altman. Altman does not have regularly scheduled meetings, but responds to messages instantly, according to his personal coach, Matt Mochary. “People inside the company get unblocked instantly and people outside the company feel totally loved” he said on a podcast. “He does this to people who are in his circle of ‘I want you to feel loved.’”

‘Think bigger’

At Y-Combinator, Altman went virtually overnight from a well-connected wunderkind to one of the best-known figures in Silicon Valley. As he transformed the incubator from scrappy start-up bootcamp into an investment powerhouse with tentacles in distant places and fields, it in turn made him a star.

“Once he took over YC,” a colleague said, “he felt like he could get a meeting with anyone.”

Altman started putting his own spin on things immediately. Less than a month after Graham named him successor, Altman put out the call for wildly ambitious founders building start-ups around breakthrough technologies, invoking Musk’s SpaceX and Tesla as examples. He listed areas like energy, AI, transportation and housing, internet infrastructure and education as of interest. “Now small start-ups can do what used to take the resources of nations,” Altman enthused on his personal blog.

“There was a profound change in the companies YC admitted after Sam became leader,” one of the people who worked with Altman said. “He used Y-Combinator as a platform to do other things, and the seeds of his demise were also there.”

Multiple people described Altman as a hands-off manager who picked potential winners and gave those people great autonomy so he could move on to his other interests. Even at OpenAI, Altman “sees himself more like an investor than a typical CEO,” said another close friend.

This style led Altman being asked to leave his role at Y-Combinator, according to four people familiar with Altman’s work there. Some perceived him as aloof and absent. He told people what they wanted to hear, said three of the people. Other leaders resented Altman “hogging credit” for building successful start-ups, the colleagues said.

Colleagues came to see Altman as off doing his own thing at OpenAI and investing in Y-Combinator companies with his own personal fund, Hydrazine. Grumbling emerged that he was reaping enormous personal profits without building and advising those companies. (Other partners also invested in YC companies on the side, according to two of the people.)

In 2019, that sentiment led to Graham flying into San Francisco from Britain, where he was living in semi-retirement after having children. People at Y-Combinator hadn’t seen Graham in years. Graham convened a short meeting with company leaders. Then Altman was asked to leave the room. Graham explained that Altman would not be returning to YC as president.

Five years after Altman took over the influential incubator, Graham said he had had no idea that Altman had been spending so little time nurturing start-ups at the organization and so much energy tending to his own projects, the people said.

Graham and his wife were “his biggest fans and his enablers,” said one of the people describing their relationship at the time. Then “they flipped.”

In an email to The Post, Graham said his wife, Jessica Livingston, a Y-Combinator founding partner, had encouraged Altman to step aside before his own visit to San Francisco, after the couple learned that he was going to be CEO of OpenAI’s new for-profit arm.

“Since he agreed immediately, it would be misleading to use the word “fired” to describe this,” he wrote. He did not respond to follow-up questions.

the first day of openai, seven years ago today pic.twitter.com/4kQUQtgb6t

— Sam Altman (@sama) January 4, 2023

A story Altman told about himself at a Y-Combinator event symbolized his bold style: Trying to win a big client, Altman flew to the company’s headquarters and sat in their lobby all day until they agreed to see him, according to someone who heard Altman tell the story. After several meetings, the leaders said they wanted to visit Altman’s offices. At the time, Altman’s firm consisted of only five people, so he hired some college friends to make his business look bigger. He said the scheme worked and he got the contract.

Altman’s tactics generate plenty of debate in the Valley. Friends say he seems to work without stress, ever busy, ever on the phone. Yet others point to the same personality traits and see a salesman who knew few limits.

“Ambition isn’t quite the right word to describe Sam,” said Groom, the close friend. “It’s something more like an inclination to say ‘Why not think bigger?’ AI companies might say, let’s try to raise 10 billion. Sam says, let’s do 100 billion — and there’s kind of a casualness about it. He has sort of an unbounded way of thinking.”

An outsider sensibility

He learned to program when he was 8. He told his brother when they played board games as children that “I have to win and I’m in charge of everything.”

Son of a real estate developer and a dermatologist, he dropped out of college to start a business. He grew up in the suburbs of St. Louis and inherited an outsider sensibility that Thiel attributed to his Jewish identity, describing Altman’s belief system to The New Yorker as “things can always go deeply wrong and that there’s no single place in the world where you’re deeply at home.”

He drives himself hard, accumulating roughly 18 pounds of muscle mass in a single year.

“I started off my career in life as a very anxious, high-strung person,” Altman told the Art of Accomplishment podcast last year. It left him, he said, “somewhat miserable..., tremendously less effective and a much worse leader.” Then he discovered meditation — his younger sister Annie, who has since cut off contact with her family, said she told Sam and her other brothers about the practice and they teased her about it, so years later, she said she was startled to find Sam had taken it on. More recently, though, Sam said on the podcast, he has mostly stopped meditating, partly because he doesn’t want to lose his motivation to work.

He and Elon Musk, the founder of Tesla and owner of what used to be Twitter, created OpenAI as a nonprofit with the aim of warning and protecting the world against a technology Musk believed could wipe out humanity by accident. Altman appeared to agree: “Development of superhuman machine intelligence is probably the greatest threat to the continued existence of humanity,” he wrote on his personal blog before the company’s launch in 2015, adding that it “does not have to be the inherently evil sci-fi version to kill us all.” But the technology’s promise was too brilliant to pass up. It just needed the right regulation, and he wanted to set up a global governing board to erect boundaries for the tool’s use.

His sister Annie said around that time her brother was more fixated on human threats, like famine or riots that could trigger violence. “People have small motivation to steal when they have housing and food,” she recalled telling him after he’d made a big purchase of guns and gold.

By the time the company launched, Altman’s rhetoric on AI risk seemed to become more modulated.

Some of Altman’s other political initiatives have remained stuck at the experimental stage. Sometimes described as a centrist Democrat, Altman from early on took public stands against Donald Trump’s 2016 presidential bid, calling Trump “abusive, erratic, and prone to fits of rage,” as well as “unfit to be president.” But Altman also credited Trump with being “right about some big things” and rebuffed calls from some tech workers for him to spurn his friend Thiel, who had become an outspoken advocate for Trump.

Some tech workers accused Altman of trying to co-opt their political movement rather than mounting any enduring effort to push back against Trump. Early in his presidency, when Trump threatened to create a registry of Muslims in the United States, many liberal tech workers signed a “Never Again” pledge, vowing not to build such a directory. Altman responded by calling an off-the-record meeting in which employees were asked to write down ideas for “Tech Worker Values,” but the initiative faded away.

That same year, Altman launched an experiment to provide a Universal Basic Income — enough money to live on, he said — to a sample group in Oakland, Calif., to see if a cash giveaway might substitute for the traditional work that AI could eliminate. “As technology continues to eliminate traditional jobs and massive new wealth gets created, we’re going to see some version of this at a national scale,” Altman wrote. “So it would be good to answer some of the theoretical questions now. Do people sit around and play video games, or do they create new things? Are people happy and fulfilled?”

Altman increasingly argued that “AI could lead to resource abundance,” meaning it could eliminate massive numbers of jobs and concentrate resources in the hands of a few, said Matt Krisiloff, a friend and former colleague of Altman’s who spearheaded the universal income project and now heads a fertility start-up that Altman invested in. “He has a belief that if one day an AI can really operate as a human — and you have factories producing robots that can do infrastructure work or farming...,” it will be “very important to figure out how those resources get distributed fairly for everyone to benefit.”

Altman said a guaranteed income would “make real progress towards eliminating poverty.” He proposed to give 100 families in Oakland $1,500 a month. When the experiment began in 2020, the terms had changed: 1,000 people were to get $1,000 a month, and a control group of 2,000 people would get $50 a month and the location had moved.

The experiment, which Altman started with $10 million, is ongoing with results expected in 2024, said Elizabeth Rhodes, research director of OpenResearch, an Altman-funded operation. Altman has not given up on the idea and has agreed not to talk about it while the experiment is ongoing.

“I would bet anything he had no idea what he was getting himself into,” Rhodes said. “He was probably thinking of self-funding a smaller thing.… In Silicon Valley there are a lot of short attention spans, but he’s stayed consistent.”

A founder ‘who can bend reality’

The consistency broke, according to some friends and critics of Altman, when he released ChatGPT, triggering the kind of corporate arms race OpenAI was founded to prevent.

Still, the controversy at OpenAI that led to Altman’s firing and rehiring last month turned not so much on the existential question of how dangerous artificial intelligence will be, but on who would control the path forward for the world’s leading AI company — a battle that has caused schisms inside the start-up for years.

Altman and Musk founded OpenAI in 2015 in part because they were worried that Google had acquired DeepMind, an AI pioneer, and seemed to be hurtling towards dominance. OpenAI recruited some of DeepMind’s talent, setting the company up as a nonprofit that they said would work for humanity’s benefit rather than financial returns.

Altman eventually took control of OpenAI after a dramatic upheaval triggered by Musk. Musk, according to two people familiar with the internal discussions, was frustrated by the lack of progress and proposed to cut half of OpenAI’s staff — a move Altman rejected. Altman, by contrast, believed OpenAI desperately needed more money to amp up its computer power and compete for talent with tech giants. Altman’s solution was to transform the company into a for-profit enterprise, though it would still be governed by a nonprofit board. The shift helped Altman secure a $1 billion investment from Microsoft.

hello from redmond! excited for the event tomorrow pic.twitter.com/b7TUr0ti42

— Sam Altman (@sama) February 6, 2023

Altman played a central role in selecting board members, said a person familiar with the board’s dealings. “Either they were his friends or they were people who can never go up against Sam without being destroyed,” the person said.

The transition to a for-profit masked internal tensions. Executives complained, to one another and openly, about Altman’s management style, calling it manipulative. “He will figure out what you want to hear,” one said. “It gets rid of the problem, but turns out to not be lasting in any way.”

Seeking to ease the friction, Altman brought in his coach, Mochary, in 2018, but employees worried he was reporting private conversations back to the CEO, according to two people familiar with the environment

Similar concerns led to the board’s decision in 2023, according to three people familiar with the proceedings. The Washington Post reported that a board review of Altman’s behavior was triggered by complaints of senior leaders alleging manipulative behavior and retaliation.

In the end, Altman’s reputation has only been burnished by his temporary downfall.

“I don’t think he is a bully,” said Khosla, who declined to comment on the chip venture. “He just asks hard questions, and sometimes people are threatened by it.”

“He’s the kind of founder that can bend reality,” said Hemant Taneja, a friend of Altman and managing director of the venture capital firm General Catalyst, adding that Altman had invited him to invest in OpenAI but that he declined because he couldn’t understand the company’s complex structure. “By creating the fastest and most popular consumer application of generative AI, he showed us the art of the possible…. This is the first technology where every CEO of every company in every industry is now thinking about how to do AI in their business. He made that happen.”

Source

“People talk about how games don’t have the emotional impact of movies. I think they do — they just have a different palette. I never felt pride, or guilt, watching a movie.”

—Will Wright, designer of The Sims

Compelling games don’t happen by accident, any more than do gripping novels, movies, or music. In all these media, creators draw from a well-defined set of strategies and techniques to create a specific emotional experience. Musicians, for instance, might use a minor key and slower tempo to create a sad or anxious mood (at least in the West). Film directors use close-ups to create intimacy. Game developers, too, have been honing techniques that lead players through intentionally designed emotional experience, whether that experience is the offbeat and comic mood of The Sims or Angry Birds, or the dramatic intensity of Call of Duty. But unlike film, fiction, or music, there isn’t yet a common language among designers, players, and society at large for what is going on and why in the hearts and minds of players.

At their heart, games differ from other media in one fundamental way: they offer players the chance to influence outcomes through their own efforts. With rare exception, this is not true of film, novels, or television. Readers and viewers of these other media follow along, reacting to the story and its twists and turns, without having a direct personal impact on the events they witness. In games, players have the unique ability to control what unfolds. As Sid Meier, designer of the best-selling game Civilization, once said, “a [good] game is a series of interesting choices.”

Actions with consequences—interesting choices—unlock a new set of emotional possibilities for game designers. Ultimately, these possibilities exist because our feelings in everyday life, as well as games, are integrally tied to our goals, our decisions, and their consequences. People go through a rapid and automatic set of evaluations as things happen to them, about what each event might mean for their goals and plans. Emotions arise in the context of these appraisals, and help guide quick and appropriate actions. Psychology researchers focused on this appraisal process, in fact, have used videogames as research instruments, in order to tightly control situations and demonstrate how particular challenges lead to emotional responses. For example, adding events that match up to someone’s in-game goals reliably induces more pride and joy in players, while adding events that block their goals leads to anger.

Researchers can actually observe the traces of meaningful choices in the brain activity of gamers.

Researchers can actually observe the traces of meaningful choices in the brain activity of gamers. Neuropsychology researchers created an experiment in which some participants played a game and others watched a live video stream of another person playing (essentially like watching a movie). The researchers used fMRI equipment to get a glimpse of everyone’s brain activity. The active gamers showed more activation of “reward-related mesolimbic neural circuits”—parts of the brain associated with motivation and reward—than those who watched passively. Interacting with the game shifted the emotional patterns observed in the players’ brains, demonstrating how we human beings experience particular rewards and emotions from the act of playing.

To the human brain, playing a game is more like actually running a race than watching a film or reading a short story about a race. When I run, I make a series of choices about actions I will take that might affect whether I win. I feel a sense of mastery or failure depending on whether I successfully execute the actions in the ways I intended. My emotions ebb and flow as I make these choices and see what happens as a result. I feel a sense of consequence and responsibility for my choices. In the end, I am to blame for the outcomes, because they arise from my own actions. This rich set of feelings that I have about the solo experience of running depends on the active role that I play in the experience—that is, on my own meaningful choices.

Flow

The ability to choose and control your actions gives rise to the second unique quality of games: the ease with which players can enter a pleasurable, optimal performance state that psychology researcher Mihaly Csikzentmihalyi calls “flow.” When people are in flow—when musicians play at their best, when athletes are in “the zone,” when programmers stay up all night creating brilliant code—time seems to melt away and personal problems disappear. Well-designed games, with the control they offer users over actions in a novel world, readily engage players in a flow state.

When Csikzentmihalyi and his colleagues studied people in flow, they isolated eight factors defining this optimal state—factors that will sound familiar to anyone who finds games compelling.

• a challenging activity requiring skill
• a merging of action and awareness
• clear goals
• direct, immediate feedback
• concentration on the task at hand
• a sense of control
• a loss of self-consciousness
• an altered sense of time

The role of the flow state in games was not lost on Csikzentmihalyi. In his book “Finding Flow,” he calls out games, “developed over the centuries for the express purpose of enriching life with enjoyable experiences,” as excellent producers of flow in human beings.

Game designers such as Jenova Chen, who did his master’s thesis work on flow, have found flow theory useful in exploring the deeper “why” behind the fun players have with games. Chen intentionally tried to reproduce the eight major factors of flow in his game design practice. By doing so, he developed some famously engaging games including Journey. Chen believes flow theory provides a working model for game designers, encouraging them to keep players in a sweet spot where they have the right amount of ability to meet the challenges at hand. Too little ability can result in anxiety and frustration; too little challenge can result in boredom or apathy.

Increasingly, game designers aim to offer players interesting choices that fall within that sweet spot, generating flow. Flow theory has been a boon to the game design and research communities, moving discussion of what players feel and why away from the vague, if positive, notion of “fun” and into more nuanced emotional territory that can be shaped through design choices. When players discuss the emotions they feel when playing games, much of their vocabulary relates to flow (curiosity, excitement, challenge, elation, or triumph) or the lack thereof (frustration, confusion, discouragement). Thus, flow theory offers a useful lens for understanding the unique emotional power of games compared to other media.

Social Emotions

When designers offer interesting choices and keep players in flow, they’re able to also start evoking another class of feelings in their players—the rich social emotions we experience in relationship with others. In the 1980s, then fledgling game company Electronic Arts (EA) released an employee recruitment advertisement that asked “Can a Computer Make You Cry?” This phrase became a rallying cry for those in the game industry interested in creating social feelings in players such as affection, camaraderie, empathy, or even grief and sadness.

This was not an unreasonable question for EA to ask, given what was known about other media. Film and novels, for example, can evoke powerful social emotional responses. People read or watch or listen, start to feel immersed in the situation being presented/described, and then feel as if they were there. They begin to care about the characters and situations as if they were real. It is not uncommon, or shameful, to cry over something happening in a film or a book.

It’s also true that over time, some viewers/readers form powerful attachments to characters, a phenomenon known as “para-social interaction.” Media creators encourage these strong feelings of connection and cultivate them through strategic design elements. For example, TV hosts may make conversational asides to viewers to generate a feeling of intimacy and familiarity. Likewise, film directors often shoot at close range to create the illusion of intimate distance between the character and the viewer. Such techniques amplify identification with the virtual people and situations.

Experts in an area of psychology called “grounded cognition” argue that these techniques evoke emotion because they mirror the way our brains make sense of the world around us in everyday life. They posit that our brains compare what we sense and experience in any given moment to our past experiences (whether “real” or “mediated”—that is, created by media) in order to come up with a set of emotional and cognitive responses that are “grounded” in experience. So if we see or hear (or form a mental picture of) a person experiencing feelings in a social setting that we, too, are immersed in, our brains are “tricked” into believing that a real social experience is taking place. Of course we engage this delusion willingly—it allows us to experience alternate situations and ways of being human, which in turn informs our own experience of being human.

This tradition, as old as oral storytelling, still provides an effective way to share emotional and social wisdom and experience. Yet in any medium other than games, we are only witnesses, not actors, and cannot affect the outcomes of the stories before us.

Grounded cognition theory helps explain what it is about games that changes the range of emotional experiences possible for players when they take on an alternate identity or social situation during play. Consider this example that led to the epigraph from Will Wright, designer of the best-selling PC game franchise of all time, The Sims. Wright describes the first time he played Black and White: Creature Isle. In this game, the player has a creature that he trains, who acts as his go-between with the villagers in the game world. The player can mold an evil creature by treating it badly, or create a moral creature by treating it kindly.

A reader or filmgoer may feel many emotions when presented with horrific fictional acts on the page or screen, but responsibility and guilt are generally not among them.

Curious about the outcome of ill treatment, Wright began to slap his creature—then was astonished to find himself feeling guilty about it, even though this was very obviously not a real being with real emotions. This capacity to evoke actual feelings of guilt from a fictional experience is unique to games. A reader or filmgoer may feel many emotions when presented with horrific fictional acts on the page or screen, but responsibility and guilt are generally not among them. At most, they may feel a sense of uneasy collusion. Conversely, a film viewer might feel joyful when the protagonist wins, but is not likely to feel a sense of personal responsibility and pride. Because they depend on active player choice, games have an additional palette of social emotions at their disposal.

Brenda Brathwaite Romero’s Train offers an elegant example of a game that calls on interesting choices and the flow state to implicate players in social choices and outcomes. Winner of the Vanguard Award at IndieCade (the primary venue for showcasing independent games in the United States), this tabletop board game is part of Romero’s Mechanics Is the Message series. (A game mechanic is an action that a player can take that changes the game state.) She created these games specifically to evoke a feeling difficult to achieve in other media—complicity.

Players of Train move boxcars full of passengers from one place to another, dealing with obstacles and challenges along the way. Only at the end of the game do they learn the train’s destination: Auschwitz. Some players realize what’s happening midgame and turn their attention to saving as many passengers as they can. Almost all players feel strong emotions after they have experienced the endgame, whether or not they realized what was going on in the midst of play. “Ultimately, I think the power of a game lies in its ability to bring us close to the subject. There is no other medium that has this power,” Romero says. “I saw people cry over Train, not just once, but multiple times. People watching, playing, those trying to save the passengers. That’s powerful, and it’s the medium that does that.”

In Train, Romero creates a tension by juxtaposing the satisfying, flow-style emotions the player feels while mastering the system and rules of the game with the negative emotions that arise from the social context of these actions. In this regard, Train can be seen as a meditation on similar painful and horrific emotional juxtapositions that may have occurred in the actual historical situation.

Romero’s other board games in the Mechanics Is the Message series are equally stark and simple, encouraging a focus on the systems and rules of games that works to forge complicity. They bring home the power of games, whether digital or not, to evoke deep, socially based emotions triggered by choice and consequence.

Source

Throw Back

In the rush to modernize and electrify vehicles, many automakers followed the lead of big tech companies and turned old-school interior knobs, switches, and buttons into touchscreens.

The only problem? Basically, lots of drivers hate them with a passion.

Apparently, enough customers in Europe complained to Volkswagen that the German car company is bringing back physical buttons and control switches to its new batch of cars, according to Autocar.

Volkswagen says future interior designs will be derived from its ID 2all concept car, an electric four-door vehicle that would cost 25,000 Euros and be sold in Europe. While it still has two screens, one for driver information and one for the infotainment system, the car features physical buttons to control it all.

Earlier this year, Volkswagen CEO Thomas Schäfer acknowledged that customers were unhappy with the vehicles' touchscreens.

"We had frustrated customers who shouldn't be frustrated. So we've spent a lot of time now – working through really systematically – on what all the functions are that a customer usually touches when using a vehicle," he told Autocar in a June interview. "We rank them. What are the most important ones? Which ones need to go on buttons? Which one needs to go on the screen? First level, second level, third level? And where do you intuitively reach when you want to switch on the light?"

Touchy

Why do drivers hate touchscreens in vehicles so much?

Navigating a screen while you're driving takes your eyes off the road. Plus they're distracting, and you have to meander through various menu options to get what you want. And functions can become inaccessible if the screen dies.

And yet touchscreens have come to dominate the market, with something like 97 percent of new vehicles now sporting these gizmos.

But customers don't necessarily like them. Consumer Reports regularly ranks cars' infotainment systems, and surveyed drivers hate the flashy, advanced systems with touchscreens or touchpads inside luxury vehicles, while significantly preferring cheaper vehicles with simpler, easier to understand controls.

It's plain to see. People want intuitive controls that are no brainers to use. They don't want to shift through multiple menu options while changing lanes. That's a recipe for disaster.

Besides Volkswagen, Rolls-Royce probably got the memo on touchscreens, and its first electric car, the Spectre, has understated interior screens and plenty of buttons and knobs that look surprisingly retro.

Hopefully, other car makers get the message: bring back physical buttons!

Source

There’s been a lot of recent chatter about artificial intelligence (AI) lately and how it could either deliver a work-free paradise or escape our control and quickly escalate into a nightmare (the likes of which have been captured in countless Hollywood blockbusters like 2001: A Space Odyssey and the Terminator franchise). Regardless of where you happen to fall on the utopia-dystopia spectrum, one thing is by now abundantly clear: AI is here to stay – and it seems almost certain to transform civilization to a degree that many can scarcely imagine.

That being the case, it’s important for all of us – including marketers, whose industry is already feeling the effects of the AI revolution – to have at least a basic understanding of what AI is and how it works. That starts with understanding some of the languages that are spoken in this strange technological territory.

Here are some critical AI terms that you need to know [we will be updating this glossary on a regular basis, so we recommend checking in on it routinely]:

A/B testing: A form of randomized experimentation wherein two variants of a particular model, A and B, are tested by a human subject to determine which of them performs better than the other.

Algorithm: A set of instructions or rules used - often by a computer - to solve a set of problems, execute calculations or process data.

AlphaGo: An AI model developed by DeepMind and designed specifically to play the ancient Chinese board game Go. In 2015, AlphaGo became the first AI model to defeat a professional human Go player (Chinese-born Fan Hui). It beat Lee Sedol, a then-professional Go player from South Korea, the following year. Sedol retired from playing professional Go in 2019, telling the South Korean media outlet Yonhap News Agency that AI specializing in the game of Go “is an entity that cannot be defeated.”

Artificial general intelligence (AGI, also sometimes referred to as Strong AI): An AI program with an intellectual ability that’s comparable to that of an average adult human. AGI, in other words, would hypothetically (we have yet to build one) be able to solve problems across a vast range of categories, just as a human brain can.

Artificial narrow intelligence (ANI, also sometimes referred to as Weak AI):  An AI program built to perform a single, narrow function, such as playing chess or responding to customer service questions. All of the AI programs that have been developed to date fall into the category of ANI.

Artificial neural network (ANN): A synthetic system, roughly modeled on the architecture of organic brains, comprised of layers of artificial neurons.

Artificial superintelligence (ASI):  First postulated by Oxford philosopher Nick Bostrum, “Superintelligence” is a theoretical intellect – artificial or organic – which is more advanced than that of humans. An ASI could have only a slightly higher IQ score than the average human being, or it could be vastly, unfathomably more intelligent, comparable to the difference in cognitive ability between an ant and Nobel Laureate Roger Penrose.

Association rule learning: A method of unsupervised and rule-based machine learning aimed at identifying commonalities or associations between variables in a dataset.

Automatic speech recognition (ASR – also known as computer speech recognition, speech-to-text or simply speech recognition): A machine’s capability to recognize human speech and then convert it into text. The iPhone dictation feature, for example, uses ASR.

Backpropagation: The process by which a neural network informs itself that it has made a predictive error, and subsequently corrects that error. The word “backpropagation” means roughly responding to flawed information by sending new information back in the direction of the source of the error. Sometimes colloquially referred to simply as “backprop” or “BP.”

Bayes’ theorem: Named after the 18th-century statistician Thomas Bayes, this theorem is a mathematical formula that can be used to determine what’s known as “conditional probability” – that is, the likelihood of a particular outcome based on one’s prior knowledge of a previous result that occurred in similar conditions.

Black box: A metaphor that’s invoked to describe a system whose inner workings are hidden and ultimately mysterious to the system’s creator (or creators). AI is sometimes described as a “black box” because models will often behave and evolve in ways that even the system’s programmers cannot fully understand or predict.

Central processing unit (CPU): The most important component of a digital computer. The CPU – sometimes referred to as the “brain” or the “control center” of a computer – is the locus of every digital computing system’s memory, arithmetic capabilities (adding, subtracting, multiplying and dividing), and the orchestrator of its operating system. The CPU of modern computers is built upon a microprocessor.

Chatbot: An AI-based computer program that leverages natural language processing (NLP) to field customer service questions in automated verbal or text-based responses that simulate human speech.

ChatGPT: An AI-powered chatbot launched by San Francisco-based startup OpenAI in November of 2022. ChatGPT uses NLP to simulate human conversation. According to OpenAI’s website, ChatGPT can “answer follow-up questions, admit its mistakes, challenge incorrect premises and reject inappropriate requests.”

Computer vision: A branch of AI that’s concerned with enabling machines to understand and respond to information derived from visual inputs - such as images and video - in a manner similar to that of the visual system in the human brain.

Convolutional neural network (CNN): A subset of artificial neural networks, commonly used in machine visual processing, which can enable an AI model to differentiate and analyze various components within an image.

Cooperative inverse reinforcement learning (CIRL): Coined by AI researchers Stuart Russell, Pieter Abbeel and others, cooperative inverse reinforcement learning (CIRL) is a hypothetical methodology for solving the so-called alignment problem, in which an AI model is designed to carry out an objective function that’s valuable to humans without knowing from the outset what that objective function is. Rather, the machine’s ability to perform the given task is enhanced through “behaviors such as active teaching, active learning and communicative actions that are more effective in achieving value alignment,” according to the paper from Russell and colleagues who first defined CIRL.

Dall-E 2: A deep learning model developed by OpenAI and released in 2022 which generates images based on the input of text-based natural language prompts. Its predecessor is Dall-E. The name of both models is a play on both the name of the title character of the Pixar film Wall-E and the surname of the 20th-century surrealist painter Salvador Dalí.

The Dartmouth Summer Research Project on Artificial Intelligence: A conference - colloquially referred to as the Dartmouth Workshop - which began in mid-1956 at Dartmouth College and is widely considered to be the event that gave birth to AI as a field of research. The conference was organized by Marvin Minsky, John McCarthy, Nathaniel Rochester and Claude Shannon.

Deep Blue: An AI program developed by IBM, the sole purpose of which is to play chess. In 1997, it made history by becoming the first intelligent machine to beat chess master Gary Kasparov in a chess match.

Deep learning (also known as deep reinforcement learning): An extension of machine learning based on the premise that machine learning models can be made more intelligent if they’re provided with vast quantities of data. Deep learning requires neural networks of at least three layers; the more layers it’s equipped with, the better its performance will be.

Deepfake: An AI-generated piece of media depicting a real person (or a real person’s voice, in the case of a deepfake audio clip). Deepfakes can be difficult to detect and are often made and spread around the internet in an effort to tarnish someone's reputation or spread some kind of misinformation.

DeepMind: An artificial intelligence research laboratory based in London and founded in 2010 by Demis Hassabis, Shane Legg and Mustafa Suleyman. The company was acquired by Google in 2014 and is now a wholly-owned subsidiary under Alphabet Inc.(Google’s parent company).

DeepMind describes itself on its website as “a team of scientists, engineers, ethicists and more, committed to solving intelligence, to advance science and benefit humanity.”

Decision tree: An imagistic illustration of the process of arriving at a decision, wherein each “branch” represents a particular course of action. Decision trees start at a “root node” (which consists of all the relevant data that’s being analyzed), branch off into “internal nodes (also known as “decision nodes”) and then terminate in “leaf nodes” (also known as “terminal nodes,” which represent all the possible outcomes of a given decision-making process).

Here’s a simple example of a decision tree rooted in the question of whether or not you should go outside to play soccer:

Emergent abilities: Unexpected patterns of behavior demonstrated by an AI model which can suddenly arise as a result of the model detecting and learning from new patterns in its training dataset. For example, a powerful large language model might develop the capacity to understand complex literary metaphors, despite the fact that such metaphors were not originally included in the model’s training data.

Entropy: In the context of machine learning, “entropy” refers to the degree of randomness, disorder and unpredictability within a dataset that’s being processed by a machine learning system. More broadly, the concept of entropy is commonly associated with the second law of thermodynamics, which essentially holds that the degree of disorder or randomness within a system will never decrease over time – it can only remain constant or increase.

Embodied agent (or “interface agent”): An entity that interacts and is reciprocally responsive with its environment through sensory-motor functions. A human being, a dog, a Boston Dynamics robot and a virtual avatar of a human are all examples of embodied agents.

Foom: An onomatopoeic word that’s supposed to represent the sound of an explosion, “foom” is used to describe a hypothetical scenario in which AI suddenly and irrevocably enters the realm of superintelligence and escapes human control.

Foundation model: A machine learning model trained on a vast corpus of unlabelled data and designed to carry out a wide variety of tasks (as opposed to a single, narrow task).

Game Theory: A mathematical formula, first postulated by mathematician John von Neumann and economist Oskar Morgenstern in 1944, relating to the dynamic interaction between two or more rational agents seeking their own gains within a parameterized (rule-governed) framework. Game Theory defines a broad set of games, including zero-sum and nonzero-sum.

Generative adversarial network (GAN): A machine learning methodology wherein two neural networks compete with one another in a zero-sum game – that is, one network’s loss translates to the other’s gain, and vice versa. Both networks are provided with a dataset, and one network called the “generator” is essentially tasked with tricking the other – the “discriminator” – into believing that the new information that’s being generated is part of the original dataset. For example, the generator might generate a new image of a human face, based on many images of real human faces, at which point the discriminator will try to determine whether or not the new image is real or manufactured. This contest will continue until the generator succeeds at tricking the discriminator with the majority (more than 50 percent) of its original output. GANs were invented in 2014 by American computer scientist Ian Goodfellow, who has henceforth been dubbed “The GANfather.”

GPT-3: Generative Pre-Trained Transformer 3 (GPT-3) is an open-source large language model developed by OpenAI and released in 2020. The model is the framework for the viral chatbot ChatGPT and is able to generate text responses in natural language based on text-based prompts.

Hallucination: In an AI context, the term “hallucination” refers to any kind of output from an AI model which is seemingly inconsistent with its training data. A hallucinating AI-powered chatbot, for example, might confidently and falsely insist that there are around 5.7tn stars in the Milky Way galaxy, even though it was not trained using any astronomical data.

Human-in-the-loop (HITL): A methodology deployed in some machine learning models wherein at least one human programmer provides feedback to the model (during testing or training) to improve the model’s performance. Ideally, HITL results in a positive feedback loop that enhances the intelligence of both machines and humans.

Hyperparameter: An overarching, predominant parameter, established by a human programmer, which determines the parameters that an AI model will establish and hone by itself during its training process.

Instrumental Convergence: A theoretical phenomenon in which all sufficiently intelligent agents – both biological and nonbiological – will ultimately identify and work towards the same instrumental goals (also called sub-goals) in pursuit of their final goals (also called absolute goals). In a hypothetical dystopian example, an AI model that’s been programmed with the final goal of removing excess carbon from the atmosphere and another that’s been given the final goal of eliminating the COVID-19 virus from the planet might determine that the best course of action is to wipe out humanity, since humans are putting huge amounts of carbon into the atmosphere and since COVID-19 is mutating inside the bodies of humans. The instrumental goals of these two models would thereby converge, even though their final goals aren't the same.

Inverse reinforcement learning (IRL): First described in a paper by AI researchers Stuart Russell and Andrew Ng, inverse reinforcement learning (IRL) is a methodology which, to put it simply, seeks to design machines that are capable of determining an agent’s (usually a human’s) goals and rewards by analyzing its behavior. An example that’s commonly invoked to describe the benefits of IRL is autonomous vehicles (AVs): Rather than trying to train an AV with every conceivable situation that it might encounter on the road - which is virtually impossible, given the fact that such possibilities are combinatorially explosive arnd therefore limitless - IRL could be leveraged to generate a comprehensive dataset of human driving behaviors and then instruct the algorithm to infer the correct course of action through a given scenario by following patterns that it detects from within that dataset.

The King Midas problem: Alluding to the Greek myth of King Midas - who had the power to turn everything he touched into solid gold, including his food and his daughter - this problem poses a crucial question related to AI research and Stuart Russel’s famous “alignment problem”: How can we be sure that an intelligent machine’s objective function is actually one which will serve the long-term best interests of human beings? In a famous thought experiment first posed by the philosopher Nick Bostrom, for example, we can imagine an AI system whose sole purpose is to create paperclips. At some point the AI decides to eliminate human beings, because they could potentially interfere with its mission, and also because it figures that the atoms in their bodies would be put to better use as raw materials for the manufacture of more paperclips. Eventually, it expands and embarks on a mindless mission to turn the entire cosmos into paperclips. The point is that we as human beings don’t always have a firm grasp on what it is that we ultimately want - and very often what we think we want ends up being bad for us.

Machine learning: A subdiscipline of artificial intelligence that, using statistical formulas and data, enables computers to progressively improve their ability to carry out a particular task or set of tasks. Crucially, a computer leveraging machine learning does not need to be explicitly programmed to improve its performance in a particular manner – rather, it’s given access to data and is designed to “teach” itself. The results are often surprising to their human creators.

Machine translation (MT): An automated process that leverages AI to translate text or speech from one language into another.

Meta-learning: In the context of AI, “meta-learning” (also sometimes described as “learning to learn”) refers to a model’s capacity to improve its ability to learn over time. Humans are also meta-learners, because we’re able to deploy a variety of strategies - such as watching and emulating others - which can gradually turn us into more effective learners, and which can improve our overall ability to navigate through the world.

Microprocessor: A CPU for digital computing systems contained within a single integrated circuit (also known as a microchip, hence the prefix in the word “microprocessor”) or a small grouping of integrated circuits. Intel introduced the world's first microprocessor, dubbed the 4004, in 1971.

Midjourney: A research lab that launched a text-to-image AI model by the same name in open beta in 2022.

Model: In the context of AI and ML, a model is an algorithm that’s trained to detect patterns and make predictions based on a particular dataset.

Model drift: The tendency for the performance of an AI model to degrade over time as its external environment changes, thereby also causing the relationship between input and output variables to change.

Moore’s Law: A principle, based on an observation usually attributed to former Intel CEO Gordon Moore, which holds that the number of transistors that can be contained within an integrated circuit (i.e., a microchip) doubles roughly every two years.

Moravec’s Paradox: Named after the computer scientist Hans Moravec, this paradox refers to the fact that machines are able to easily carry out functions that are difficult for most human beings – such as performing complex mathematical calculations – and yet they struggle to do things – like perform basic motor tasks or read social cues – that most human beings are able to do with little to no effort.

Natural language processing (NLP): A branch of artificial intelligence – that also blends elements of linguistics and computer science – aimed at enabling computers to understand verbal and written language in a manner that imitates the human brain’s language-processing capability.

Objective function: A mathematical formula that's used to measure the accuracy of an AI model’s predictions, and thereby its overall ability to carry out a given task. An objective function can be likened to a numerical score which shows an AI model how effectively it’s performing; a high score conveys that it’s on the right track, while a low score suggests that it probably needs to implement a different problem-solving strategy.

OpenAI: A non-profit AI research lab founded in 2015 by Sam Altman, Elon Musk and others. As its name suggests, the original foundational goal of OpenAI was to collaborate with other organizations in the field of AI and to open-source its research. In 2019, the organization launched a “capped profit” subsidiary called OpenAI Limited Partnership (OpenAI LP). (Musk has lamented this decision on Twitter.)

Parameter: A variable within the process of training an AI model which can be adjusted by the model in order to hone its ability to produce a particular output using a given dataset.

Pattern recognition: An automated process whereby a computer is able to identify patterns within a set of data.

Prior probability (also sometimes referred to simply as a prior): A term used in the field of Bayesian statistics to refer to the assigned likelihood of an event before (prior to) additional (posterior) information necessitates the revision of that likelihood.

Reinforcement learning (RL): The process of teaching machine learning models to make optimal decisions within a dynamic environment. When using RL, a programmer will often present a machine learning model with a game-like situation in which one outcome is preferable to others. The machine then proceeds to experiment with different strategies and the programmer will “reinforce” the desired behavior with rewards and discourage other behaviors through penalties.

Reinforcement learning from human feedback (RLHF), or reinforcement learning from human preferences: Like traditional reinforcement learning (RL), Reinforcement learning from human feedback (RLHF) aims to fine-tune the performance (or the “policy”) of a machine learning model by rewarding (i.e., reinforcing) certain actions. Unlike RL, however, RLHF involves human beings in the process of choosing which of the model’s actions should be rewarded. OpenAI’s ChatGPT, for example, was trained using RHLF.

Self-supervised learning: A branch of machine learning wherein an AI model is provided with unlabeled data and is allowed to label the data according to its own pattern recognition capabilities. A self-supervised algorithm will then use those initial labels as it continues to interpret subsequent iterations of data input.

Semi-supervised learning: A branch of machine learning which, as the name suggests, blends elements of both supervised learning and unsupervised learning. Semi-supervised learning is based on the input of some labeled data and a higher quantity of unlabeled data, the goal being to teach an algorithm to categorize the latter into predetermined categories based on the former, and also to allow the algorithm to identify new patterns across the dataset. It is widely considered to be a kind of bridge, connecting the benefits of supervised learning with those of unsupervised learning.

Supervised learning: A branch of machine learning based on the input of clearly labeled data and aimed at training algorithms to recognize patterns and accurately label new data.

Stochastic: A mathematical term referring to a system’s tendency to produce results that are unpredictable. (Roughly synonymous with “probabilistic,” “indeterminable” and “random.”) Many AI algorithms are programmed to incorporate some degree of randomness into their learning processes and are therefore described as stochastic. The results of a deterministic system, in contrast, can reliably be predicted beforehand.

TensorFlow: An open-source platform, developed by Google, designed for the management of machine learning and AI systems.

Transformer: First defined in a 2017 paper from Google Brain researchers, a transformer is a neural network which learns to identify and understand the contextual relationships within datasets, thus enabling it to produce new, cogent outputs. ChatGPT, for example, is based upon a transformer model which specializes in understanding the contextual relationships between individual words. Through such a contextual understanding, it’s able to predict which word should logically follow those that preceded it. (“GPT” stands for “generative pre-trained transformer.”)

Turing test: A blinded experiment – invented by and named after 20th-century mathematician Alan Turing – where a human subject interacts with an artificially intelligent machine and asks it a series of questions. If the human interlocutor is unable to say definitively whether the responses are being generated by a human or an AI, the latter has passed the Turing Test.

Uncanny valley: A theoretical concept, first postulated by roboticist Masahiro Mori in 1970, which refers to an eerie, uncanny quality that will be perceived by a human being interacting with an artificial entity that closely (though imperfectly) resembles another human.

Unsupervised learning: A branch of machine learning which is based upon the input of unlabeled data. In contrast to supervised learning, unsupervised learning allows an algorithm to create its own rules for identifying patterns and categorizing data.

Value alignment problem: Coined by computer scientist Stuart Russel, the phrase “value alignment problem” – or simply “alignment problem” – refers to the difficulties that come with ensuring that intelligent machines share the same values and goals as their human programmers. This problem has spawned a subfield of AI and machine learning called “alignment research.”

Weights: Parameters which neural networks autonomously learn to optimize in order to more accurately detect patterns in datasets and make predictions. Weights are the artificial counterparts to synapses, which are responsible for the transmission of electronic signals between neurons in organic brains.

Source

A Wisconsin family has agreed to sell 407 acres of their land, which includes a local pumpkin farm attraction, to Microsoft — for a total of $76 million, reported the Milwaukee Business Journal. The local government initially offered the Creuziger family in Mount Pleasant, Wisconsin about a third of that sum in 2017 as a part of an agreement with the Foxconn Technology Group. But the family refused, opting to hold out for a better offer.

“The family wishes the village and Microsoft well, and they would appreciate people respecting their privacy,” the family’s attorney David Barnes told the Business Journal. 

The land, which also includes the Land of the Giants pumpkin farm and a nine-acre corn maze, neigbhors another 641 acres of land purchased by Microsoft from the village of Mount Pleasant — for a total of $99.7 million. The end goal for Microsoft is to build a data center campus in the area, in which it plans to invest over $1 billion. 

All told, the sale is a happy ending for the village, after several years of confusion and mixed signals from Foxconn. It’s only been two years since Foxconn drastically scaled back its promised $10 billion investment in Mount Pleasant, which would have included a state-of-the-art manufacturing plant.

The Milwaukee Business Journal reported that Microsoft plans to initially hire 200 employees at its Mount Pleasant data center — and could add over 460 jobs over time. But it’ll still be a fraction of the 13,000 jobs that Foxconn originally promised the area back in 2017.

Source

A few days ago, Intel CEO Pat Gelsinger raised a few eyebrows by attributing Nvidia's current dominance of the AI market to luck. He argued that if Intel hadn't killed its supercomputing GPU project in the 2000s, it would be where Nvidia is now. When Intel ditched the project, it also jettisoned Gelsinger, so he might be a bit salty about how things turned out. An engineer who worked on Intel's defunct "Larrabee" project but now works at Nvidia has responded to Gelsinger, saying it wasn't luck but Nvidia's vision and execution, which Intel lacks. Ouch.

The employee doing the smack talk is Bryan Catanzaro, whose Twitter bio says he is VP of Applied Deep Learning Research at Nvidia. We've seen him before in videos talking about the company's Deep Learning Super Sampling technology, or DLSS. He notes in his tweet that he was one of the engineers working on Larabee in 2007, and his LinkedIn profile shows he was an intern at Intel back then.

We previously wrote that Intel claimed Larrabee could do ray tracing in modern games at acceptable frame rates, so it was ahead of its time. Also, it seems there's some notable crossover between Intel's hopes for Larrabee in 2007 and what Nvidia produced with Turing in 2018.

Back then, there was an actual skirmish between Nvidia and Intel on the HPC processing front, with Nvidia promoting its Tesla cards and Intel promoting its Knights Ferry GPGPUs. We wrote then, "Having seen lackluster adoption of Nvidia's Tesla platform in HPC applications, Intel hopes that this level of cross-compatibility will ensure Intel's continued dominance in supercomputing."

Things did not turn out in Intel's favor, as Intel essentially killed the Larrabee project—choosing instead to go with its x86 processors for HPC tasks and focusing its graphics efforts on integrated GPUs. When Intel officially announced Larrabee in 2010, AnandTech wrote that Intel's decision "completely validates Nvidia’s Tesla strategy."

Catanzaro notes in his tweet that Intel was 10 times the size of Nvidia back then. Therefore, it figured it would crush Nvidia with Larrabee. Still, it lacked the vision and execution to see the project through to what Nvidia is doing now with its advantages in parallel computing. A quick Google search backs up these numbers, as in 2007, Nvidia's annual revenue was roughly $3 billion, while Intel's was $38 billion.

Fast forward to Q3 of 2023, and the tables have turned. Nvidia's earned $18 billion for the quarter, a 206% increase from a year ago. Meanwhile, Intel earned $14 billion, a decrease of 8% year-over-year. More important was where the companies made those dollars, with Nvidia pulling in $14.5 billion from its data center products, while Intel earned just $3.8 billion.

It remains to be seen if this war of words will continue. We doubt it will go on much longer as it doesn't make Gelsinger look particularly charitable toward one of Intel's partners. He noted in his previous comments that he's good friends with Nvidia's CEO and that they talk often, making us wonder if this topic has ever come up in their discussions.

Source

Key notes

• Wikipedia is once again asking for donations to help it survive.
• The website is asking for $2.75 from each reader in the United States.
• Elon Musk has previously criticized the platform for asking too much donation.

Wikipedia, the free online encyclopedia, is once again asking for donations to help it survive. The website is asking for $2.75 from each reader in the United States. 

 

Facing financial difficulties, Wikipedia has implemented a pop-up message requesting a $2.75 donation from each reader. This has been going on for quite some time, but if you open Wikipedia today, you’ll likely be greeted with this big banner:

“In the age of AI, access to verifiable facts is crucial. Wikipedia matters more than ever as a reliavle source for emerging technologies, and you. Your gift supports how readers use Wikipedia now, and how revolutionary new systems will utilize it tomorrow,” the message reads.

 

Wikipedia is one of the most popular websites in the world, amassing over 4.5 billion monthly global visits according to Statista. The website is funded by donations from readers and by grants from foundations since its beginning in 2001. 

 

Elon Musk, the owner of X (fka Twitter) among other things, and the world’s wealthiest man, has openly feuded with the non-profit before beefing with Google Bard. He questioned the site’s constant appeal for donations, spitting a rather childish joke that he would give $1 billion if they changed it into “Dickipedia.” Eeks. 

 

Source

When the Clay Mathematics Institute put individual $1-million prize bounties on seven unsolved mathematical problems, they may have undervalued one entry—by a lot. If mathematicians were to resolve, in the right way, computer science’s “P versus NP” question, the result could be worth worlds more than $1 million—they’d be cracking most online-security systems, revolutionizing science and even mechanistically solving the other six of the so-called Millennium Problems, all of which were chosen in the year 2000. It’s hard to overstate the stakes surrounding the most important unsolved problem in computer science.

P versus NP concerns the apparent asymmetry between finding solutions to problems and verifying solutions to problems. For example, imagine you’re planning a world tour to promote your new book. You pull up Priceline and start testing routes, but each one you try blows your total trip budget. Unfortunately, as the number of cities grows on your worldwide tour, the number of possible routes to check skyrockets exponentially, rapidly making it infeasible even for computers to exhaustively search through every case. But when you complain, your agent writes back with a solution sequence of flights. You can easily verify whether or not their route stays in budget by simply checking that it hits every city and summing the fares to compare against the budget limit. Notice the asymmetry here: finding a solution is hard, but verifying a solution is easy.

The P versus NP question asks whether this asymmetry is real or an illusion. If you can efficiently verify a solution to a problem, does that imply that you can also efficiently find a solution? Perhaps a clever shortcut can circumvent searching through zillions of potential routes. For example, if your agent instead wanted you to find a sequence of flights between two specific remote airports while obeying the budget, you might also throw up your hands at the similarly immense number of possible routes to check, but in fact, this problem contains enough structure that computer scientists have developed a fast procedure (algorithm) for it that bypasses the need for exhaustive search.

You might think this asymmetry is obvious: of course one would sometimes have a harder time finding a solution to a problem than verifying it. But researchers have been surprised before in thinking that that’s the case, only to discover last-minute that the solution is just as easy. So every attempt in which they try to resolve this question for any single scenario only further exposes how monumentally difficult it is to prove one way or another. P versus NP also rears its head everywhere we look in the computational world well beyond the specifics of our travel scenario—so much so that it has come to symbolize a holy grail in our understanding of computation.

In the subfield of theoretical computer science called complexity theory, researchers try to pin down how easily computers can solve various types of problems. P represents the class of problems they can solve efficiently, such as sorting a column of numbers in a spreadsheet or finding the shortest path between two addresses on a map. NP represents the class of problems for which computers can verify solutions efficiently. Our book tour problem, called the Traveling Salesperson Problem by academics, lives in NP because we have an efficient procedure for verifying that our agent’s solution worked.

Notice that NP actually contains P as a subset because solving a problem outright is one way to verify a solution to it. For example, how would you verify that 27 x 89 = 2,403? You would solve the multiplication problem yourself and check that your answer matches the claimed one. We typically depict the relationship between P and NP with a simple Venn diagram:

Credit: Amanda Montañez

The region inside of NP but not inside of P contains problems that can’t be solved with any known efficient algorithm. (Theoretical computer scientists use a technical definition for “efficient” that can be debated, but it serves as a useful proxy for the colloquial concept.) But we don’t know if that’s because such algorithms don’t exist or we just haven’t mustered the ingenuity to discover them. Here’s another way to phrase the P versus NP question: Are these classes actually distinct? Or does the Venn diagram collapse into one circle? Do all NP problems admit efficient algorithms?

Here are some examples of problems in NP that are not currently known to be in P:

• Given a social network, is there a group of a specified size in which all of the people in it are friends with one another?
• Given a varied collection of boxes to be shipped, can all of them be fit into a specified number of trucks?
• Given a sudoku (generalized to n x n puzzle grids), does it have a solution?
• Given a map, can the countries be colored with only three colors such that no two neighboring countries are the same color?

Ask yourself how you would verify proposed solutions to some of the problems above and then how you would find a solution. Note that approximating a solution or solving a small instance (most of us can solve a 9 x 9 sudoku) doesn’t suffice. To qualify as solving a problem, an algorithm needs to find an exact solution on all instances, including very large ones.

Each of the problems can be solved via brute-force search (e.g., try every possible coloring of the map and check if any of them work), but the number of cases to try grows exponentially with the size of the problem. This means that if we call the size of the problem n (e.g., the number of countries on the map or the number of boxes to pack into trucks), then the number of cases to check looks something like 2n. The world’s fastest supercomputers have no hope against exponential growth. Even when n equals 300, a tiny input size by modern data standards, 2300 exceeds the number of atoms in the observable universe. After hitting “go” on such an algorithm, your computer would display a spinning pinwheel that would outlive you and your descendants.

Thousands of other problems belong on our list. From cell biology to game theory, the P versus NP question reaches into far corners of science and industry. If P = NP (i.e., our Venn diagram dissolves into a single circle) and we obtain fast algorithms for these seemingly hard problems, then the whole digital economy would become vulnerable to collapse. This is because much of the cryptography that secures such things as your credit card number and passwords works by shrouding private information behind computationally difficult problems that can only become easy to solve if you know the secret key. Online security as we know it rests on unproven mathematical assumptions that crumble if P = NP.

Amazingly, we can even cast math itself as an NP problem because we can program computers to efficiently verify proofs. In fact, legendary mathematician Kurt Gödel first posed the P versus NP problem in a letter to his colleague John von Neumann in 1956, and he expressed (in older terminology) that P = NP “would have consequences of the greatest importance. Namely, it would obviously mean that ... the mental work of a mathematician concerning yes-or-no questions could be completely replaced by a machine.”

If you’re a mathematician worried for your job, rest assured that most experts believe that P does not equal NP. Aside from the intuition that sometimes solutions should be harder to find than to verify, thousands of the hardest NP problems that are not known to be in P have sat unsolved across disparate fields, glowing with incentives of fame and fortune, and yet not one person has designed an efficient algorithm for a single one of them.

Of course, gut feeling and a lack of counterexamples don’t constitute a proof. To prove that P is different from NP, you somehow have to rule out all potential algorithms for all of the hardest NP problems, a task that appears out of reach for current mathematical techniques. In fact, the field has coped by proving so-called barrier theorems, which say that entire categories of tempting proof strategies to resolve P versus NP cannot succeed. Not only have we failed to find a proof but we also have no clue what an eventual proof might look like.

Source

Although Google has been busy producing generative AI tools to sell as part of its ever-evolving cloud business model, a new report by The Information suggests the company may soon start using AI itself, threatening some workers' jobs.

After months of studies and reports claiming that AI will ‘help’ humans and not ‘replace’ them, the biggest fear affecting many workers could have more body than previously thought.

According to the most recent report, the company told workers that it would be reorganizing its 30,000-person ad sales unit, leaving many fearing about more redundancies to come.

Is Google replacing workers with AI?

Already this year, the company has laid off more than 12,000 workers as a result of rising costs caused by global economic uncertainties. In the following months, Google’s AI business started to take off, and it has now situated itself as one of the leaders in artificial intelligence along with OpenAI and Microsoft.

The company is now reportedly looking at using AI, ML, and automation to drive its ads business. It has already been injecting GenAI into the ad experience for advertisers and customers, but this new report indicates that the time-saving technology may soon be used on the back end, too.

Lots of advertisers have already started using the Performance Max ad tool, which was updated earlier this year, meaning that fewer workers could now be needed to specialize in certain ad topics.

The report reveals that last year, 13,500 of the company’s 30,000-strong ad sales unit were working on projects that have been simplified or replaced by Performance Max.

It’s unclear how many workers would be affected if any at all, but with such boosts to efficiency, it’s clear that unless Google receives a surge in customers, it may not need so many workers.

TechRadar Pro has asked Google to comment on the matter and confirm whether its AI has had an effect on job security. Any update will be posted here.

Source