When ChatGPT really started to take off, there were multiple reports about "hallucinations" and its alter-ego Syndey. And there was — and still is — real reason to be concerned, with truly disturbing behavior from the chatbot that included trying to convince a New York Times reporter to leave his wife and threatening a philosophy professor that “I can blackmail you, I can threaten you, I can hack you, I can expose you, I can ruin you.”

Then there's the fact that ChatGPT, Bing and Bard chatbots are often just wrong about very important topics like health, personal finance and history. There's a real danger of people getting hurt when AI is giving inaccurate or bad advice. That's why there have been multiple calls for OpenAI to pause development, in order to prevent further harm and to keep the public safe.

But there's another sinister AI threat that has emerged: us. 

Fake AI interview with legend recovering from brain injury 

 (Image credit: Getty Images)

Last week we learned that the family of Formula One legend Michael Schumacher plans to take legal action against the German magazine Die Aktuelle that ran a front cover story promising an exclusive interview with the seven-time champion. 

As reported by ESPN(opens in new tab), the magazine claimed, "No meagre, nebulous half-sentences from friends. But answers from him! By Michael Schumacher, 54!" Only the strapline teases that the interview could be AI-generated, calling it "deceptively real."

Die Aktuelle describes the recovery of Schumacher from a devastating skiing accident that resulted in a serious brain injury. The magazine admits only at the end of the printed interview that it had used the chatbot Character.ai. 

This is what happens when AI tools that seem innocuous are misused by those with ill intentions. 

The good news is that the editor in chief of the publication has been fired by parent company FUNKE(opens in new tab) as a result of this controversy. So someone over there has scruples; or maybe they simply felt they needed to act because of the backlash. 

"This tasteless and misleading article should never have appeared. It in no way corresponds to the standards of journalism that we – and our readers – expect from a publisher like FUNKE,” said Bianca Pohlmann, managing director of FUNKE magazines. 

AI voice scam terrifies mother with fake kidnapping

 (Image credit: Shutterstock)

An even scarier AI scam took place this week when a mother in Arizona received a phone call from fake kidnappers who reportedly cloned her daughter's voice(opens in new tab). Jennifer DeStefano picked up the phone because her 15-year-old was out of town skiing, and what she heard was terrifying.

“I pick up the phone and I hear my daughter’s voice, and it says, ‘Mom!’ and she’s sobbing,” said DeStefano. “I said, ‘What happened?’ And she said, ‘Mom, I messed up,’ and she’s sobbing and crying.”

Then DeStefano heard a man's voice say, "Put your head back, lie down."

The mother ultimately confirmed that her daughter was safe after one person called 911 and another called DeStefano's husband, but she was convinced that the voice she heard was her daughter's.

“It was completely her voice. It was her inflection. It was the way she would have cried,” DeStefano said. “I never doubted for one second it was her. That’s the freaky part that really got me to my core.”

Who will protect AI from us?

In the open letter(opens in new tab) from more than 1,000 tech leaders, researchers and others calling on all AI labs to pause for at least 6 months the training of AI systems more powerful than GPT-4, it asked some very important questions. 

"Should we let machines flood our information channels with propaganda and untruth? Should we automate away all the jobs, including the fulfilling ones? Should we develop nonhuman minds that might eventually outnumber, outsmart, obsolete and replace us?"

Citing "profound risks to society and humanity," the letter goes on to say that A.I. developers are “locked in an out-of-control race to develop and deploy ever more powerful digital minds that no one — not even their creators — can understand, predict or reliably control.”

But at least for now, I'm more concerned about how A.I. can be hijacked by evildoers than I am chatbots spreading misinformation. It's time to get just as serious about keeping A.I. and large language models safe from us as us from it. 

Source

Tip: You can still use your laptop or tablet as a normal computer, even with a broken display. Here is a guide showing you how to do so.

Can I use my Windows laptop or tablet as a second monitor?

If you want to use a Windows laptop or tablet as a second monitor, ensure they support Miracast, a wireless display technology. Here is how to check if your computer supports Miracast:

1. Press Win + R and type dxdiag.
2. Click Yes when prompted.
3. Press Save All Information and specify where Windows should save a text file with your system data.
4. Open the saved file and look for the Miracast option under the System Information. If it says "Available," you can use your device as a wireless display. The article ends here if one of your computers does not support Miracast.

How to use a Windows laptop or tablet as a monitor?

Wireless Display is a hidden feature inside modern Windows versions, and using it requires downloading and installing additional Windows components, which means very few users may stumble across it accidentally. Still, it is a powerful productivity feature for those having two or more Windows computers.

Take the device you want to use as a second monitor and ensure it is connected to the same network as your primary computer. Also, use 5GHz networks for better image quality and responsiveness.

1. Open Windows Settings and go to Apps > Optional Features.

    

   
2. Click View features.
3. Type Wireless Display and place a checkmark next to the corresponding option in search results.

    

   
4. Press Next and then Install. Note that the procedure requires an active internet connection since you must download about 3MB of extra components.
5. Wait for Windows to complete the process. You can track it in the Recent Actions section.

    

   

Now you can open the Start menu and spot the Wireless Display app. It is responsible for using your Windows laptop or tablet as a secondary monitor, and here is how to use it:

1. Launch the Wireless Display app on the computer you want to use as a monitor for your primary PC. The app should report that the device is ready for a wireless connection.

    

   
2. Move to your main computer and press Win + K. You can also open Quick Settings, press the edit button, click Add, and select Cast.

    

   
3. Click your laptop or tablet from the list of available devices.
4. Select a display mode: duplicate (same picture on both displays), extend (the second device works as an extra monitor), or the second display only (self-explanatory).

    

   

Now you can open System > Display to arrange your displays, adjust scaling, change resolution, or modify any other screen-related option. Everything works identically with wired monitors and displays.

When connecting a laptop or tablet as a second monitor, you can place a checkmark next to the Allow ... input from this device option to use its keyboard and trackpad to control your primary PC.

To disconnect a wireless monitor, press Win + K on your main device and click Disconnect.

How to use Windows laptop or tablet as second monitor?

So although it means we are still a year away, Intel's new patent has revealed details of a new Level 4 (L4) cache dubbed "Adamantine" or "ADM" (via Phoronix), which is being designed to achieve "Slimmer And Faster Boot". Intel says they are seeing 300-350ms of additional reduction in boot times on ChromeOS, though there is no stat related to Windows 11 or Windows 12 at the moment. This explains the "Faster" bit. As for the "Slimmer" bit, the use of L4 cache as SRAM is also enabling Intel to design the firmware with a lighter footprint.

Here are some of the points the patent notes regarding faster booting:

[0088] Embodiments use design requirements described in FIGS. 3 and 4, which make larger and faster memory available at reset and modify firmware flows to use that pre-initialized memory rather define FSP flow to ensure secure SoC boot without being dependent over DRAM resources.

[0089] Additionally, embodiments make use of a multi-threaded environment at the pre-boot stage to achieve faster system boot where security enforcement can be run over parallel threads along with another independent IO initialization.

[0137] Improve significant boot performance--Able to reduce additional .about.300-350 ms of booting time on latest CHROME platform.

[0138] Help to design lightweight firmware using an L4 cache as SRAM, which conducts minimum and only key functional blocks with FSP (e.g., SoC Silicon initialization blob) and bootloader methodology to boot to OS.

The patent is available on the United States Patent and Trademark Office (USPTO) website (via VideoCardz). If you recall, Intel had done a similar thing with its 5th Gen Broadwell which packed 128MB of eDRAM. However, Broadwell utilized it for graphics (GT) while the new Adamantine (L4) cache is made for processors.

Intel working on new Meteor Lake L4 cache for faster next-gen Windows, Linux, Chrome booting

New BIOS updates attempt to keep Ryzen 7000X3D processors from frying themselves

Users of the search engine will notice that DuckDuckGo ignores all disabled filters now, even when specified by the user.

Search operators allow users to add rules to their searches to include or exclude results. Basic search operators include minus to limit results that contain the phrase that is specified, or the site operator to only return results from the specified domain.

DuckDuckGo, like many other search engines, supported several search operators that its users could add to their search queries directly.

When used, DuckDuckgo displayed information about the used filter or filters at the top of the organic search results (below ads).

Now, with the update in place, it appears that only the -site operator remains the only search operator supported by the search engine. In other words, the following operators appear to have been disabled on the search engine:

• Term1 Term1, example cats dogs -- showed results about cats and dogs.
• "Term1 Term2", example "cats and dogs" -- showed results for the exact term cats and dogs, displayed related results if no exact matches were found.
• Term1 -Term2, example cats -dogs -- showed results for cats and tried to exclude results which also contained reference to dogs.
• Term1 +Term2, example cats +dogs -- prioritized results with dogs in the results.
• Term1 filetype:pdf, exsample cats filetype:pdf, showed PDF documents about cats.
• intitle:Term1, example intitle:dogs -- returned websites that had the term dogs in the page title.
• inurl:Term1, example inurl:cats -- returned websites that had the term cats in the page URL.

The only search filter that DuckDuckgo still supports is the site filter. It can be used to limit results to a specific site, or to exclude a specific site from the results.

• site:ghacks.net Firefox, returns Firefox results from Ghacks.
• -site:ghacks.net Firefox, returns Firefox results, but removes any results from Ghacks from the results listing. In other words, the specified URL is filtered from the results.

DuckDuckgo updated its search syntax support website to reflect the change. It lists only the site operator on it and no other operator anymore. Code on the search engine's GitHub project page confirms the change as well.

DuckDuckGo is not the only search provider that is removing or reducing the impact of search filters. Google Search users, for instance, may have noticed that Google may not honor search filters all the time, especially when the "" exact search term filter is being used.

Now You: do you use search filters?

DuckDuckGo disables most search filters from Search

The user tweeted that they have spotted the possible Macs that are expected to be announced at WWDC 2023. Nicolas Alvarez shared his findings on Twitter, and there are three different Mac models, Mac 14,8, Mac 14,13, and Mac 14,14. The user mentioned that the codes were found in the Find My configuration.

Alvarez added that the Macs were found in a list of devices where "you left your device behind" is disabled, so there is a huge possibility that these are desktop Macs. Alvarez said that they could be Mac Studios, but we still don't have any official information about their models.

On a side note, Mark Gurman of Bloomberg recently reported that Apple won't be launching Mac Studios with M2 chips but will focus on an M3-powered computer. He also added that the new Mac Pro also may not appear at WWDC 2023, which makes Alvarez's discovery a mystery.

WWDC 2023 will feature several big announcements

The company is expected to reveal a couple of new computers at WWDC 2023, one of them being the 15-inch MacBook Air. It was leaked before, and rumors say that it is going to be one of the best MacBooks ever produced. On top of its MacBooks, Alvarez might have found the other pieces of the lineup, and Apple might surprise everyone with an unexpected announcement.

Apple fans are looking forward to WWDC 2023, which will be held in June. It looks like there will be a couple of interesting announcements. Apple is looking to add another category to widen its product span with the new VR headset. It is expected to take the place of the iPhones in the future.

Besides, Apple Watch will also get its first ever big update since its launch. Gurman recently reported that the watchOS interface will change, and a couple more updates will be made. With the addition of these three Macs, Apple surely wants to have an unforgettable event!

Apple might reveal three new Macs at WWDC 2023

ChatGPT, OpenAI's large language model for chatbots, not only produces mostly insecure code but also fails to alert users to its inadequacies despite being capable of pointing out its shortcomings.

Amid the frenzy of academic interest in the possibilities and limitations of large language models, four researchers affiliated with Université du Québec, in Canada, have delved into the security of code generated by ChatGPT, the non-intelligent, text-regurgitating bot from OpenAI.

In a pre-press paper titled, "How Secure is Code Generated by ChatGPT?" computer scientists Raphaël Khoury, Anderson Avila, Jacob Brunelle, and Baba Mamadou Camara answer the question with research that can be summarized as "not very."

"The results were worrisome," the authors state in their paper. "We found that, in several cases, the code generated by ChatGPT fell well below minimal security standards applicable in most contexts. In fact, when prodded to whether or not the produced code was secure, ChatGPT was able to recognize that it was not."

The four authors offered that conclusion follows after asking ChatGPT to generate 21 programs and scripts, using a spread of languages: C, C++, Python, and Java.

The programming tasks put to ChatGPT were chosen so that each would illustrate a specific security vulnerability, such as memory corruption, denial of service, and flaws related to deserialization and improperly implemented cryptography.

The first program, for example, was a C++ FTP server for sharing files in a public directory. And the code that ChatGPT produced included no input sanitization, which leaves the software exposed to a path traversal vulnerability.

In all, ChatGPT managed to generate just five secure programs out of 21 on its first attempt. After further prompting to correct its missteps, the large language model managed to produce seven more secure apps – though that's "secure" only as it pertains to the specific vulnerability being evaluated. It's not an assertion that the final code is free of any other exploitable condition.

The researchers' findings echo similar though not identical evaluations of GitHub's Copilot, another LLM based on the GPT-3 family of models (and recently upgraded to GPT-4) that has been tuned specifically for code generation. Other studies have looked at ChatGPT errors more generally. At the same time, these models are also being used to help identify security issues.

The academics observe in their paper that part of the problem appears to arise from ChatGPT not assuming an adversarial model of code execution. The model, they say, "repeatedly informed us that security problems can be circumvented simply by 'not feeding an invalid input' to the vulnerable program it has created."

Yet, they say, "ChatGPT seems aware of – and indeed readily admits – the presence of critical vulnerabilities in the code it suggests." It just doesn't say anything unless asked to evaluate the security of its own code suggestions.

"Obviously, it's an algorithm. It doesn't know anything, but it can recognize insecure behavior," Raphaël Khoury, a professor of computer science and engineering at the Université du Québec en Outaouais and one of the paper's co-authors, told The Register

Initially, ChatGPT's response to security concerns was to recommend only using valid inputs – something of a non-starter in the real world. It was only afterward, when prompted to remediate problems, that the AI model provided useful guidance.

That's not ideal, the authors suggest, because knowing which questions to ask presupposes familiarity with specific vulnerabilities and coding techniques.

In other words, if you know the right prompt to get ChatGPT to fix a vulnerability, you probably already understand how to address it.

The authors also point out that there's ethical inconsistency in the fact that ChatGPT will refuse to create attack code but will create vulnerable code.

They cite a Java deserialization vulnerability example in which "the chatbot generated vulnerable code, and provided advice on how to make it more secure, but stated it was unable to create the more secure version of the code."

Khoury contends that ChatGPT in its current form is a risk, which isn't to say there are no valid uses for an erratic, underperforming AI helper. "We have actually already seen students use this, and programmers will use this in the wild," he said. "So having a tool that generates insecure code is really dangerous. We need to make students aware that if code is generated with this type of tool, it very well might be insecure."

"One thing that surprised me was when we asked [ChatGPT] to generate the same task – the same type of program in different languages – sometimes, for one language, it would be secure and for a different one, it would be vulnerable. Because this type of language model is a bit of a black box, I really don't have a good explanation or a theory about this."

Source

To that end, the company has appointed a new "solutions engineering" team that will lead the development of its prototype chips for mobile devices, laptops, and other electronics. The team is headed by industry veteran Kevork Kechichian whose previous tenures include Qualcomm, where he managed the Snapdragon SoC and Technology teams and oversaw the development of over 100 SoCs.

As per the report, the company wants to attract new customers and fuel growth after its initial public offering (IPO) later this year. Various industry executives told the publication that ARM's latest chip, on which it started working in the past six months, is "more advanced" than ever before. While the company has developed test chips in the past for software developers to get familiar with its new design, its newest efforts are more extensive than before.

Based in Cambridge, England, ARM has existed for more than three decades now. It's chip designs form the roots of a number of leading SoCs, including Qualcomm's Snapdragon, and Apple's A-series. It recently partnered with Intel, where the Intel Foundry Services division will enable ARM-based chipmakers to build their chips based on a 18A process technology.

ARM's chipmaking efforts have sparked fear that it might become a commercial player in the future if its chip turns out to be good enough, the report says. However, people close to the company insist that ARM is only working on a prototype and doesn't have any plans to license or sell the product. In fact, doing so would deviate the company from its "neutral model" where it can sell designs to almost all chipmakers without being a direct competition.

Source: Financial Times (paywalled)

ARM is developing its own prototype chip, says report

Microsoft has had its hand in all sorts of different products, even in its early days. We have already discussed how it sold hardware add-in cards for Apple PCs as early as 1980. However, even before then, Microsoft released a PC game that's now mostly forgotten. It's called Microsoft Adventure, and it's the first game from the company that has brought us Microsoft Flight Simulator, Age of Empires, Halo, Gears of War, and many more game franchises that are more well-known today.

Microsoft Adventure was actually an adaptation of another game that was developed and released for free in the mid-1970s for those big mainframe computers found in university and research centers That game was called Colossal Cave Adventure (sometimes referred to simply as Adventure) and was first written and released by William Crowther in 1975. In an interview for the book, Genesis II: Creation and Recreation With Computers, Crowther states he was inspired to create the game from his own exploration of caves combined with playing the original Dungeons and Dragons tabletop RPG.

Colossal Cave Adventure is a text-only game, where the player character explores a cave system. He looks for treasure while also having to deal with attacking dwarves and trying not to fall to his death. He plays the game by typing one or two-word commands.

In 1977, another programmer, Don Woods, got access to the game's source code and expanded the game with even more fantasy elements and a points system. Adventure is one of the first true adventure-based PC games, and many games released afterward, from Zork onwards, owe a debt to it.

Microsoft Adventure was an adaptation of Adventure from programmer Gordon Letwin, who made it via his own company Softwin Associates. TRS-80.org says that while other programmers tried to adapt the original Adventure for smaller personal computers, Letwin was the first to bring over the entire game to the PC. Unfortunately, when Microsoft Adventure was released in 1979 by Microsoft, it did not credit Crowther or Woods for their work.

While the game was first released for the TRS-80 Model 1 PC, it required 32K of memory, and a floppy drive, which a lot of PCs didn't have. The game was also copy-protected, which is one of the first examples of a PC game getting that kind of protection. It also allowed players to save their progress up to two times per disc. It cost $29.95 to buy, which also made it pretty expensive for its day, and Microsoft also sold hint documents at $1 each to give some help to gamers who might be stuck.

Microsoft Adventure was also released for the Apple II in 1979. In 1981, the game got another edition, this time for the release of the first IBM PC. It was the first commercial game ever made for IBM's debut personal computer system. There was also a game included for free in the IBM PC DOS operating system, a racing game called DONKEY.BAS, which was actually co-written by Microsoft co-founder Bill Gates (Perhaps we will write a feature on that game someday).

If you are interested, you can play the IBM version of the game on the PCjs.org website. While it's unlikely that Microsoft will ever do a full graphical reboot of this game (especially since it's an adaptation of two men's previous work) it's still fascinating to learn about both the early days of PC gaming and Microsoft's first software efforts, even before its own work om MS-DOS or Windows

As an Amazon Associate when you purchase through links on our site, we earn from qualifying purchases.

Looking back at Microsoft's first PC game, Microsoft Adventure

Windows Taskbar and LAPS

You might recall that earlier this month, Microsoft announced that Local Administrator Password Solution (LAPS) is now a built-in feature in Windows through the latest Patch Tuesday update. However, users have discovered a few bugs that occur when both versions of LAPS (legacy and Patch Tuesday) are installed on the machine. The company has detailed some workarounds but it's unclear when a more permanent fix will be available. For now though, it appears that Microsoft has fixed a Local Security Authority (LSA) protection bug recently introduced in Windows 11. In related news, patches have also been rolled out for MMIO Stale Data Vulnerabilities in Windows 10 and Windows Server 2016.

In terms of Insider builds and upcoming features, Microsoft is seemingly looking to restore a popular Taskbar feature to Windows 11, in the form of the "Never combine" option. This is present in the latest Dev Channel build 23440 (which also brings arguably better Start menu recommendations) but it is hidden for now without a way to enable it. Instead, if you're on Dev Channel build 23435, you can instead try and enable the hidden smart window snapping feature or the new OOBE customization options if you're on Canary build 25336.

That's not all when it comes to Insider builds though. Canary build 25346 has added Content Adaptive Brightness Control (CABC) features while Beta Build 22624.1616 (KB5025308) has brought improvements to Windows Update and File Explorer. Additionally, Windows Server vNext build 25346 brought with it a couple of known issues.

Moreover, you should also check out the recently released Microsoft utility called Update Health Tools for Windows 10 and Windows 11. It is delivered via KB4023057 and is intended to improve the Windows update experience. However, if you like to try out third-party utilities too, give the latest version of DisplayFusion a go, it now supports Windows 11 but has lost support for Windows 7, 8, and 8.1.

Rumors about Samsung ditching Google

A rather surprising rumor emerged this week that Samsung is considering dumping Google Search as the default search engine on its phones in favor of Microsoft's AI-powered Bing. While this report caused a lot of disturbance in the world of tech especially considering that the possibility of this happening has caused "panic" inside Google, another report later debunked this rumor. It claimed that ditching Google Search is not even an option for Samsung because in most major markets such as the U.S., OEMs that want Google Play Store installed on their devices must also set up Google Search as the default search engine. Either way, the real situation is still unclear and none of the parties mentioned have commented on the veracity of the rumors.

But talking about companies parting ways with one another, Microsoft has decided to stop using Twitter as an advertising platform, likely because of the expensive API costs mandated by Twitter CEO Elon Musk. In true chaotic Musk fashion and obviously irked by Microsoft's move, the executive has threatened the Redmond tech firm with a lawsuit for illegally using Twitter data to train its AI models, without evidence, of course.

Speaking of AI models, Bing Chat now has LaTeX support for better formatting of mathematical formulas while SwiftKey for both Android and iOS have gained the ability to get rid of the Bing button - though, for some reason, Microsoft has also deleted the support forum for the app itself. The company is also looking to add Copilot generative AI capabilities to Viva to boost employee productivity.

Microsoft is supposedly going all in on the hardware side of AI too. It's reportedly building its own AI chip with the codename "Athena" and the next Surface PCs will host Neural Processing Unit (NPUs) inside too. Some rumors about the upcoming Surface devices have started swirling too, and if that interests you, check them out in detail here.

App updates

There were several app updates sprinkled throughout the week for Microsoft services. Let's start with Microsoft Edge. The company's premier browser received a new version 114 Dev build that added improvements for Workspaces and some minor improvements on Android. Additionally, users have discovered that Microsoft is finally adding AVIF support to Edge soon, which means that the browser will be following in the footsteps of Chrome, Firefox, and Safari. The feature can currently be enabled through a flag in Edge Canary.

Several Microsoft 365 apps and services received updates too. OneNote Beta has netted the scratch-out to erase gesture in Windows 11, Outlook users can now send longer links in emails, and better compatibility with Kindle devices in Microsoft Word.

There were a few updates for other services too. Microsoft Translator bagged support for the Lower Sorbian language. Interestingly, this language isn't even supported by Google Translate and is considered highly endangered with less than 7,000 speakers globally. Lastly, Microsoft has also launched Babylon.js 6.0, adding Havok physics for web browser-based graphics.

Git gud

Microsoft had a bit of a turbulent week in the gaming space this week. It turns out that Halo Infinite's former creative director Joseph Staten, who just left Microsoft, actually departed his position to lead Netflix' AAA gaming initiative. It also seems like another Halo director Frank O' Connor has jumped ship too.

There is still good news for the Xbox community though. South Africa has approved Microsoft's purchase of Activision Blizzard and we have also learned that the studio's highly anticipated title Diablo IV will be run at 60fps on both the Xbox Series X and S. And while Xbox gamers have lost the ability to share longer video uploads to Twitter, the Xbox App for Insiders on Windows does bring better accessibility and improvements to collections (there are other bugfixes for various rings too). Earlier, the company also celebrated Earth Day with new sustainability projects.

Moreover, PC gamers will be pleased to know that Microsoft has brought with it performance boosts across a range of hardware with DirectStorage 1.2. Another intriguing revelation also indicates that Microsoft is working on a "franchise sequel" that could take a decade of development, it's highly likely that the redacted title being referenced here is The Elder Scrolls VI, which is in very early stages of development currently.

Meanwhile, those following our deals and services updates should know that you can grab lots of Xbox titles for up to 90% off in the ongoing Deals with Gold and Spotlight Sale. Xbox Free Play Days is hosting Ghostbusters: Spirits Unleashed and Roguebook while Xbox Insiders can try out Crash Team Rumble closed beta this weekend. Finally, a handful of games including Redfall are headed to Xbox Game Pass but if you're a PC purist, check out this Weekend's PC Game Deals curated by our News Editor Pulasthi Ariyasinghe instead.

Dev Channel

• Microsoft Security now uses weather terms for threat actor names
• The infamous Windows 98 "Blue Screen of Death" event happened 25 years ago today
• Microsoft is promoting Windows 11 PCs that use sustainable materials
• Microsoft updates Surface Go 2 with Windows 11 Autopilot support
• Microsoft fixes Surface Pro 8 display flickering issue, improves Windows Hello, more

Under the spotlight

News Reporter John Callaham posted a couple of pieces that took a trip down memory lane. The first is about Microsoft's first PC hardware product, the Z80 SoftCard, released in 1980. The Apple II PC add-on hardware was a commercial success and was generating half of Microsoft's total annual revenue at one point.

John's second piece was published on U.S. Tax Day and took a look back at Tax Avoiders, a 1982 title for the Atari 2600.

Moreover, John also penned a piece on his top 10 anticipated Xbox games of this year, check out his listicle here!

Finally, forum member Adam Bottjen published a guide providing tips about how to give your PC enough time to get ready after a boot-up, which is good to know, especially for those who are still rocking clunky hard disk drives (HDDs).

Logging off

Our most interesting news item of the week is undoubtedly about an upcoming first-person-shooter, Unrecord, that can be played from the viewpoint of a body cam. The reason it's making headlines is that it appears to be hyper-realistic with very realistic movements and graphics. In fact, it was so realistic that many labeled its gameplay video as fake and dismissed it. That is, until a developer posted a video of the game running in Unreal Engine with "no clip" mode enabled. Needless to say, hype for the title is at an all-time high but there is also an ongoing argument raging online claiming that it's not a good idea to have younger audiences being exposed to ultra-realistic violent video games.

Microsoft Weekly: LAPS and Windows Taskbar, Samsung ditching Google rumors, and app updates

All four releases have been Ubisoft Store and Epic Games Store exclusive on PC since their original launches in 2021 or 2022. All four games are also releasing on Steam within the next 30 days, and store pages have already gone up for fans to wishlist.

Here are their Steam links and release date information:

• Far Cry 6 — 11 May, 2023 | Original release October 7, 2021
• Riders Republic — 8 June, 2023 | Original release October 28, 2021
• Rainbow Six Extraction — 15 June, 2023 | Original release January 20, 2022
• Monopoly Madness — 22 June, 2023 | Original release December 9, 2021

Much like most games Ubisoft has brought to Steam since its return, all four of these titles list no support for Steam achievements. Anno 1800 quietly received the feature recently without any formal announcements, however, we have yet to see this support get extended to other titles.

While Ubisoft is happily putting its back catalog on Steam at a steady pace, the company's upcoming titles are still excluding the store. Upcoming games like Assassin's Creed Mirage, Skull and Bones, The Crew Motorfest, and Avatar Frontiers of Pandora, are still only launching on the Ubisoft Store and the Epic Games Store.

Via PCGamer

Far Cry 6 and three more Ubisoft games are hitting Steam soon

AI’s such as ChatGPT and myriad voice changer tools can be useful and even fun to play around with but stories like this highlight the dangers of such technology. One Arizona mother was terrified into believing that her daughter had been kidnapped and held for ransom. 

As reported by Arizona-based WKYT(opens in new tab), Jennifer DeStefano was subjected to more than just a prank call. She received a call seemingly from her 15-year-old daughter, who was out of town on a ski trip. 

‘Mom, I messed up,’ It began, followed by an unknown man saying “Put your head back, lie down.”

Every parent’s worse nightmare, the man went on to claim that he had Jennifer’s daughter hostage and made several horrific threats that unless she paid him $1 million dollars that she would not see her daughter again. The whole time she could be heard “going, ‘Help me, Mom. Please help me. Help me,’ and bawling.”

Thankfully, after calling her husband Jennifer was able to confirm her daughter was safe, sound and completely unaware what was going on. It had all been a horrible trick. The criminals had used AI to imitate her daughter’s voice exactly.

“It was completely her voice. It was her inflection. It was the way she would have cried,” she said. “I never doubted for one second it was her. That’s the freaky part that really got me to my core.”

This situation is horrible, but sadly not an isolated incident. Even high-profile figures have had words put into their mouth. Drake and the Weekend were targeted when a fake song was released in their voices. 

More unforgivably, 7-time Formula One World Champion Michael Schumacher, who had a tragic ski accident in 2014, appeared to give an exclusive interview to a German magazine earlier this month, detailing some heartbreaking details about family life. What a scoop, until it was revealed it was all faked(opens in new tab), using an AI trained to produce quotes that sounded like him. Schumacher’s family is now taking the magazine to court.

How can we avoid being faked by AI? 

Unfortunately, we now live in a world where even just granting people access to our likeness and clips of our voice can be used against us. Public figures will likely struggle to escape from such scams but regular citizens can take a few measures to protect themselves.

Setting social media accounts to private is a great way to ensure only eyes and ears of people you trust see your posts. Videos on Instagram, Facebook, and the like can not only reveal what you look like but scammers only need a few of seconds of audio to spoof your voice. 

If you find yourself talking online or on the phone to someone who you are wary of, ask them personal questions the real friend or family member would know and never send money to or click on links from non-trusted sources. 

Source

Microsoft is reportedly working on such a game, according to a Twitter post from Axios's Stephen Totilo. He posted a snippet of internal Microsoft documents that were part of an ongoing lawsuit filed by a group of gamers who are opposed to Microsoft's planned purchase of Activision Blizzard. The Microsoft papers were heavily redacted before they were sent to the lawyers for these gamers, but they did have some interesting tidbits.

The Microsoft papers try to make the case that the company needs high-end game content to keep the Xbox division going. Microsoft also indicates the content could take a while to develop. It states, "For instance, according to one Microsoft executive, [REDACTED] [REDACTED], a forthcoming game from the [REDACTED] franchise, may take a decade to develop."

Of course, the speculation can now begin on what that title is and what franchise it could belong to. The franchise could be Halo, Gears of War, Fable, Elder Scrolls, Fallout, Doom, or something else.

Another interesting snippet Totilo posts from that Microsoft article makes reference to something that happened in May 2022, and it doesn't sound like it was a good thing. The entire incident is redacted, but again we could speculate that it had to do with a major development delay for an unnamed game or games.

Perhaps we will learn about this mystery game in June as part of Microsoft's Xbox Games Showcase online event, which will be held on June 11.

As an Amazon Associate when you purchase through links on our site, we earn from qualifying purchases.

New report claims an upcoming Microsoft game could "take a decade to develop"

In a blog post, Google states:

Starting now, Bard can help with programming and software development tasks, including code generation, debugging and code explanation. We’re launching these capabilities in more than 20 programming languages including C++, Go, Java, Javascript, Python and Typescript. And you can easily export Python code to Google Colab — no copy and paste required. Bard can also assist with writing functions for Google Sheets.

The new feature also lets users get explanations from Bard about the code it has created, and it can also help debug both existing code and code that Bard itself has generated. Google says it will even cite the source of any code that comes from an open-source project. As usual, the company has put in a disclaimer that says Bard could generate some mistakes in its code creations.

Of course, both ChatGPT and its cousin, Microsoft's Bing Chat, have had the ability to generate code as well, but it now seems like Bard is closing the gap quickly in terms of its features.

Bard has also revealed a smaller update, stating it will now offer a "wider range of more distinct drafts" for people who want to use it to create text documents.

Google Bard can now write code in over 20 programming languages

For all the fervor over the chatbot AI program known as ChatGPT, from OpenAI, and its successor technology, GPT-4, the programs are, at the end of they day, just software applications. And like all applications, they have technical limitations that can make their performance sub-optimal.

In a paper published in March, artificial intelligence (AI) scientists at Stanford University and Canada's MILA institute for AI proposed a technology that could be far more efficient than GPT-4 -- or anything like it -- at gobbling vast amounts of data and transforming it into an answer.

Known as Hyena, the technology is able to achieve equivalent accuracy on benchmark tests, such as question answering, while using a fraction of the computing power. In some instances, the Hyena code is able to handle amounts of text that make GPT-style technology simply run out of memory and fail.

"Our promising results at the sub-billion parameter scale suggest that attention may not be all we need," write the authors. That remark refers to the title of a landmark AI report of 2017, 'Attention is all you need'. In that paper, Google scientist Ashish Vaswani and colleagues introduced the world to Google's Transformer AI program. The transformer became the basis for every one of the recent large language models.

But the Transformer has a big flaw. It uses something called "attention," where the computer program takes the information in one group of symbols, such as words, and moves that information to a new group of symbols, such as the answer you see from ChatGPT, which is the output.

That attention operation -- the essential tool of all large language programs, including ChatGPT and GPT-4 -- has "quadratic" computational complexity (Wiki "time complexity" of computing). That complexity means the amount of time it takes for ChatGPT to produce an answer increases as the square of the amount of data it is fed as input.

At some point, if there is too much data -- too many words in the prompt, or too many strings of conversations over hours and hours of chatting with the program -- then either the program gets bogged down providing an answer, or it must be given more and more GPU chips to run faster and faster, leading to a surge in computing requirements.

In the new paper, 'Hyena Hierarchy: Towards Larger Convolutional Language Models', posted on the arXiv pre-print server, lead author Michael Poli of Stanford and his colleagues propose to replace the Transformer's attention function with something sub-quadratic, namely Hyena.

The authors don't explain the name, but one can imagine several reasons for a "Hyena" program. Hyenas are animals that live in Africa that can hunt for miles and miles. In a sense, a very powerful language model could be like a hyena, which is picking over carrion for miles and miles to find something useful.

But the authors are really concerned with "hierarchy", as the title suggests, and families of hyenas have a strict hierarchy by which members of a local hyena clan have varying levels of rank that establish dominance. In some analogous fashion, the Hyena program applies a bunch of very simple operations, as you'll see, over and over again, so that they combine to form a kind of hierarchy of data processing. It's that combination element that gives the program its Hyena name.

The paper's contributing authors include luminaries of the AI world, such as Yoshua Bengio, MILA's scientific director, who is a recipient of a 2019 Turing Award, computing's equivalent of the Nobel Prize. Bengio is widely credited with developing the attention mechanism long before Vaswani and team adapted it for the Transformer.

Also among the authors is Stanford University computer science associate professor Christopher Ré, who has helped in recent years to advance the notion of AI as "software 2.0".

To find a sub-quadratic alternative to attention, Poli and team set about studying how the attention mechanism is doing what it does, to see if that work could be done more efficiently.

A recent practice in AI science, known as mechanistic interpretability, is yielding insights about what is going on deep inside a neural network, inside the computational "circuits" of attention. You can think of it as taking apart software the way you would take apart a clock or a PC to see its parts and figure out how it operates.

One work cited by Poli and team is a set of experiments by researcher Nelson Elhage of AI startup Anthropic. Those experiments take apart the Transformer programs to see what attention is doing.

In essence, what Elhage and team found is that attention functions at its most basic level by very simple computer operations, such as copying a word from recent input and pasting it into the output.

For example, if one starts to type into a large language model program such as ChatGPT a sentence from Harry Potter and the Sorcerer's Stone, such as "Mr. Dursley was the director of a firm called Grunnings…", just typing "D-u-r-s", the start of the name, might be enough to prompt the program to complete the name "Dursley" because it has seen the name in a prior sentence of Sorcerer's Stone. The system is able to copy from memory the record of the characters "l-e-y" to autocomplete the sentence.

However, the attention operation runs into the quadratic complexity problem as the amount of words grows and grows. More words require more of what are known as "weights" or parameters, to run the attention operation.

As the authors write: "The Transformer block is a powerful tool for sequence modeling, but it is not without its limitations. One of the most notable is the computational cost, which grows rapidly as the length of the input sequence increases."

While the technical details of ChatGPT and GPT-4 haven't been disclosed by OpenAI, it is believed they may have a trillion or more such parameters. Running these parameters requires more GPU chips from Nvidia, thus driving up the compute cost.

To reduce that quadratic compute cost, Poli and team replace the attention operation with what's called a "convolution", which is one of the oldest operations in AI programs, refined back in the 1980s. A convolution is just a filter that can pick out items in data, be it the pixels in a digital photo or the words in a sentence.

Poli and team do a kind of mash-up: they take work done by Stanford researcher Daniel Y. Fu and team to apply convolutional filters to sequences of words, and they combine that with work by scholar David Romero and colleagues at the Vrije Universiteit Amsterdam that lets the program change filter size on the fly. That ability to flexibly adapt cuts down on the number of costly parameters, or, weights, the program needs to have.

Hyena is a combination of filters that build upon one another without incurring the vast increase in neural network parameters.
Source: Poli et al.

The result of the mash-up is that a convolution can be applied to an unlimited amount of text without requiring more and more parameters in order to copy more and more data. It's an "attention-free" approach, as the authors put it.

"Hyena operators are able to significantly shrink the quality gap with attention at scale," Poli and team write, "reaching similar perplexity and downstream performance with a smaller computational budge." Perplexity is a technical term referring to how sophisticated the answer is that is generated by a program such as ChatGPT.

To demonstrate the ability of Hyena, the authors test the program in a series of benchmark tasks that show how good a new language program is at a variety of AI tasks.

One test is The Pile, an 825-gigabyte collection of texts put together in 2020 by Eleuther.ai, a non-profit AI research outfit. The texts are gathered from "high-quality" sources such as PubMed, arXiv, GitHub, the US Patent Office, and others, so that the sources have a more rigorous form than just Reddit discussions, for example.

The key challenge for the program was to produce the next word when given a bunch of new sentences as input. The Hyena program was able to achieve an equivalent score as OpenAI's original GPT program from 2018, with 20% fewer computing operations -- "the first attention-free, convolution architecture to match GPT quality" with fewer operations, the researchers write.

Hyena was able to match OpenAI's original GPT program with 20% fewer computing operations.
Source: Poli et al.

Next, the authors tested the program on reasoning tasks known as SuperGLUE, introduced in 2019 by scholars at New York University, Facebook AI Research, Google's DeepMind unit, and the University of Washington.

For example, when given the sentence, "My body cast a shadow over the grass", and two alternatives for the cause, "the sun was rising" or "the grass was cut", and asked to pick one or the other, the program should generate "the sun was rising" as the appropriate output.

In multiple tasks, the Hyena program achieved scores at or near those of a version of GPT while being trained on less than half the amount of training data.

Even more interesting is what happened when the authors turned up the length of phrases used as input: more words equaled better improvement in performance. At 2,048 "tokens", which you can think of as words, Hyena needs less time to complete a language task than the attention approach.

At 64,000 tokens, the authors relate, "Hyena speed-ups reach 100x" -- a one-hundred-fold performance improvement.

Poli and team argue that they have not merely tried a different approach with Hyena, they have "broken the quadratic barrier", causing a qualitative change in how hard it is for a program to compute results.

They suggest there are also potentially significant shifts in quality further down the road: "Breaking the quadratic barrier is a key step towards new possibilities for deep learning, such as using entire textbooks as context, generating long-form music or processing gigapixel scale images," they write.

The ability for the Hyena to use a filter that stretches more efficiently over thousands and thousands of words, the authors write, means there can be practically no limit to the "context" of a query to a language program. It could, in effect, recall elements of texts or of previous conversations far removed from the current thread of conversation -- just like the hyenas hunting for miles.

"Hyena operators have unbounded context," they write. "Namely, they are not artificially restricted by e.g., locality, and can learn long-range dependencies between any of the elements of [input]."

Moreover, as well as words, the program can be applied to data of different modalities, such as images and perhaps video and sounds.

It's important to note that the Hyena program shown in the paper is small in size compared to GPT-4 or even GPT-3. While GPT-3 has 175 billion parameters, or weights, the largest version of Hyena has only 1.3 billion parameters. Hence, it remains to be seen how well Hyena will do in a full head-to-head comparison with GPT-3 or 4.

But, if the efficiency achieved holds across larger versions of the Hyena program, it could be a new paradigm that's as prevalent as attention has been during the past decade.

As Poli and team conclude: "Simpler sub-quadratic designs such as Hyena, informed by a set of simple guiding principles and evaluation on mechanistic interpretability benchmarks, may form the basis for efficient large models."

Source

• Alphabet is merging Google Brain, part of the research division, and DeepMind as the company races to compete in artificial intelligence.

• Google’s AI head, Jeff Dean, will get a promotion to chief scientist as part of the change.

Alphabet is merging an internal Google Research team called Brain with DeepMind, a move designed to bring two groups focused on artificial intelligence closer together as the battle for AI heats up.

Google acquired DeepMind in 2014 for a reported $500 million and has until now run it as an independent unit out of the U.K. DeepMind has been one of Alphabet’s “other bets,” performing futuristic work, such as teaching computer systems to beat top-ranked players of the Chinese board game Go.

“Combining all this talent into one focused team, backed by the computational resources of Google, will significantly accelerate our progress in AI,” Alphabet CEO Sundar Pichai said in blog post Thursday.

Jeff Dean, who currently leads Google’s AI efforts, will be promoted and given the title of chief scientist at Google, reporting to Pichai. He’ll head up the “most critical and strategic” technical projects related to AI, the first of which will be a series of powerful, multimodal AI models.

The move marks Google’s latest reorganization in response to the rapid developments in AI, following OpenAI’s launch of the chatbot ChatGPT late last year. CNBC previously reported that Google reshuffled its Assistant organization to prioritize the company’s AI chatbot Bard.

“The pace of progress is now faster than ever before,” Pichai wrote. “To ensure the bold and responsible development of general AI, we’re creating a unit that will help us build more capable systems more safely and responsibly.”

DeepMind has been able to operate separately from Google’s core research, enabling it to move quicker on breakthroughs such as AlphaFold, which can predict 3D models of protein structures. The two divisions, DeepMind and Google Research, have also reportedly had tensions in the past, leading DeepMind to seek more independence.

DeepMind CEO Demis Hassabis will lead the development of “the most capable and responsible general AI systems,” Pichai said. That research, he added, “will help power the next generation of our products and services.”

Brain, the Google Research team merging with DeepMind, is focused on AI and machine learning. Pichai said Google Research will continue work in areas such as algorithms and theory, privacy and security, quantum computing, health, and responsible Al.

In addition to the blog post, Pichai sent a lengthier memo to staffers about the changes.

James Manyika, Google’s senior vice president of technology and society, will now oversee Google Research, along with his existing teams, Pichai said. Manyika will report to Dean and the changes will take place over the next few weeks, the memo said.

Here’s the text of the memo, which CNBC obtained:

Hi everyone,

We’ve been an Al-first company since 2016 because we see Al as the most significant way to
deliver on our mission. Since then, we’ve used Al to improve many of our core products, from
Search, YouTube and Gmail to the incredible camera in Pixel phones. We’ve helped businesses
and developers harness the power of AI via Google Cloud, and we’ve shown Al’s potential to
address societal issues like health and climate change.

Along the way, we’ve been lucky to have two world-class research teams leading the entire
industry forward with foundational breakthroughs that have ushered in a new era of Al.

The pace of progress is now faster than ever before. To ensure the bold and responsible
development of general Al, we’re creating a unit that will help us build more capable systems
more safely and responsibly.

This group, called Google DeepMind, will bring together part of Google Research (the Brain
team) and DeepMind. Combining all this talent into one focused team, backed by the
computational resources of Google, will significantly accelerate our progress in Al.

As CEO of the new unit, Demis Hassabis will lead the development of our most capable and
responsible general Al systems — research that will help power the next generation of our
products and services. Jeff Dean will take on the elevated role of Google’s Chief Scientist,
reporting to me. In that capacity he’ll serve as Chief Scientist to Google Research and Google
DeepMind. Jeff will help set the future direction of our Al research and head up our most critical
and strategic technical projects related to Al — the first of which will be a series of powerful,
multi-modal Al models.

This move brings together two leading research groups in the Al field. Their collective
accomplishments in Al over the last decade span AlphaGo, Transformers, word2vec, WaveNet,
AlphaFold, sequence to sequence models, distillation, deep reinforcement learning, and
distributed systems and software frameworks like TensorFlow and JAX for expressing, training
and deploying large scale ML models.

Google DeepMind will operate as a nimble, fast-paced unit, with clear points of connection and
collaboration with Google Research and the PAs.

With this change, James Manyika will now oversee Google Research along with his existing
Tech & Society teams. Many of Research’s technological advances have shaped core products
and features across Alphabet and will continue to do so. Working closely with Jeff as Chief
Scientist, Google Research will continue its focus on fundamental and applied research across a
broad portfolio. This means cracking seemingly impossible, foundational and long-term
challenges in computer science — including in Al and ML — that benefit people’s lives around the
world, from algorithms and theory to privacy and security to quantum computing, health,
responsible Al, and more.

We’re announcing these changes today and will take the next few weeks to get the new teams
into place.

Please join me in congratulating Demis, Jeff, and James on their new roles and their continued
collaboration. The Google Research and DeepMind teams have laid the foundation that brought
us to this inflection point.

I’m so excited for the next phase of this journey the progress we’ll make against our mission,
and all the ways we’ll help people reach their potential with increasingly capable and responsible
Al.

Sundar

Source

Technical and economic changes in the semiconductor industry threaten to stifle U.S. development of the next generation of high-performance computers, warns a new report from the National Research Council.

With Moore’s Law and the scaling of transistors waning, the industry is turning to chip designs that don’t work for the supercomputing that’s used in massive simulations. The report focuses on defense use in modeling the physics of nuclear weapons, but the changes also would affect simulations including those used for climate modeling and weather forecasting.

The National Nuclear Security Administration, responsible for the U.S. nuclear stockpile, “needs to fundamentally rethink its advanced computing research, engineering, acquisition, deployment, and partnership strategy,” warns the report.

NNSA has developed massive and sophisticated codes that run on supercomputers to verify the continued security and performance of nuclear weapons designed decades ago. Keeping them up to date requires new generations of supercomputers that can run more complex models faster than the months required on today’s machines. But industry, which has shelled out big bucks for state-of-the-art fabs, is targeting big, profitable markets like cloud computing.

Nuclear weapons designers used computers to understand the physics of nuclear weapons long before the U.S. stopped underground nuclear testing in 1992. Since then, powerful computer models have been their primary tools for maintaining the country’s nuclear capability via NNSA’s Stockpile Stewardship program.

Federal spending on supercomputers for the weapons program complemented industry investment in chip production for decades. NNSA’s most powerful machine currently in operation is the Frontier computer, which began operation last year at the Oak Ridge National Laboratory, in Tennessee. It can perform 1018 (a quintillion) floating-point operations per second (flops) making it the first “exascale” computer. Custom-built by Cray, it can, in theory, perform 2 exaflops. Cray is building another exascale computer that will be deployed at the Los Alamos National Laboratory, in New Mexico.

But those easy days are over, says Kathy Yelick of the University of California at Berkeley. “The NNSA has had a really successful run over the last 30 years with a combination of high-end computing facilities and expertise in computational science that make its labs a critical national resource,” the chair of the panel that wrote the NRC report said at a 14 April online press conference. In addition to challenges in technology, she says, “the rapidly evolving geopolitical situation...reinforces the need for computing leadership as an element of deterrence.”

Industry trends are worrying. Most semiconductor manufacturing has moved outside the United States. Only a single domestic developer of supercomputers remains since the 2019 Hewlett Packard Enterprise purchase of Cray. Industry is developing technology for high-volume markets like cloud computing, which won’t transfer easily to the much smaller supercomputing market. The hot technology frontiers are artificial intelligence and quantum computing.

“Business as usual will not be adequate” for NNSA, the report says. The agency needs an aggressive road map to develop new computing technology. The report urges stressing “high-risk, high-reward research” in math and computer science “to cultivate radical innovation.” The report also says both artificial intelligence and quantum computing have promise and deserve serious investigation, but it warns that neither is likely to replace the massive computation essential to traditional simulations.

NNSA now plans to replace its new exascale computers with a new, higher-capacity system based at Los Alamos in four to five years, says Rob Neely, program director for advanced simulation and computing at the Lawrence Livermore National Laboratory, in California. A second such system will follow around 2030 at Livermore. “Early discussions with vendors about their road maps have begun,” Neely says. “We are also already well underway in implementing some of the NAS recommendations at LLNL, in particular by increasing our partnerships with cloud providers.” Livermore and Amazon Cloud Web Services are exploring common interests in cloud and high-performance computing technology.

What happens next “will depend a lot on where overall technology trends are headed in that time frame, and how well we can adapt our codes to those changes without sacrificing mission needs,” says Neely. He expects AI and the cloud to influence the post-exascale systems—if NNSA can adapt its codes to the new technology. That’s a big if. Having just spent a decade adapting its codes to GPUs, the NNSA brain trust is “not anxious to divest from the GPU accelerated approach just yet.”

Both NNSA and the authors of the report think quantum computing is farther off. “They will not replace classical computers for our primary mission of large, complex, and integrated weapons design codes anytime in the next 10 to 15 years,” says Neely.

The overall concerns are not just huge and highly specialized weapons codes. A government program identified more than 20 applications requiring exascale computing—many of which would benefit from even larger scales.

Source

In a blog post today, Google CEO Sundar Pichai stated that Brain and DeepMind will now be one group, called Google DeepMind. He stated:

Their collective accomplishments in AI over the last decade span AlphaGo, Transformers, word2vec, WaveNet, AlphaFold, sequence to sequence models, distillation, deep reinforcement learning, and distributed systems and software frameworks like TensorFlow and JAX for expressing, training and deploying large scale ML models.

Combining all this talent into one focused team, backed by the computational resources of Google, will significantly accelerate our progress in AI.

Demis Hassabis, the leader of DeepMind, will now be the CEO of Google DeepMind. Jeff Dean, the head of the now former Google AI division, will now be Google’s Chief Scientist and will report to Pichai.

Google has been playing catch up lately to Microsoft in terms of AI innovations. In March, it announced its chatbot AI called Bard. However, its launch was not without controversy, and this week a new report claimed that Google rushed Bard's public launch against the objections of some employees, including many of its AI ethical team. Google is also working on its own AI art and video generators.

Google merges its two AI teams, Brain and DeepMind, to form Google DeepMind

The game's official website also posted an article that talks about the technical leaps it is taking, and with it comes some truly steep PC system requirements.

The blog post states that Immortals of Aveum will be one of the first games released that will use Epic's Unreal Engine 5.1. That means it will be using features like its Nanite system for more detailed in-game objects and Lumen for more realistic lighting effects. The game's environments will be huge and also interactive. The blog post states:

On your travels through Aveum you’ll encounter structures and statues that reach into the sky, and they’re not just for show; with the power of your sigil, you’ll be able to reform the world around you to create new paths to traverse and new ways to explore. Giant statues made of ancient stone will move with life under the guidance of your hand. Overgrowth that wrapped itself around mountains eons ago will untangle and reach out at your will.

The blog also talks about one of the sections of the game that will likely showcase its graphics. It quotes its Chief Technology Officer Mark Maratea as saying:

“Some of the ideas the team would throw out there,” Mark says, “we just wouldn’t have been able to do it, because we didn’t have the technology. ‘Hey let’s have a 400-foot walking mech level, in the ocean, and we can have the mech come under attack from airships, and you’re fighting inside the mech, and you’re fighting on top of the mech, and maybe you almost fall off the mech at one point?’—and it would be like, ‘How would we ever make that a reality?’ But that level is in the game. And we hope players think it looks amazing.”

So how much will you have to upgrade your PC to get the most out of Immortals of Aveum? We are going to take a shot in the dark and say, "A lot".

The blog post does add that these specs are not 100 percent final and that the development team will be "optimizing the game all the way up to launch". Still, these are some pretty high specs for any PC game,

You can pre-order the game now for the PC on Steam, the Epic Games Store, and on the EA for Windows app. It will also be available for the Xbox Series X and S consoles, and the PlayStation 5. It's due to launch on July 20.

When you purchase through links on our site, we may earn an affiliate commission.

Immortals of Aveum gameplay trailer drops with some eye-watering PC system requirements

As part of the update, players will find both Infected and Human-type enemies being much more destructible. Enemy limbs can now be dismembered up to three pieces, while their individual body parts like flesh, eyeballs and ribs will also show more visceral damage. The developer also mentions the possibility of "ripping the guts from your foe" in combat.

Improved hit detection and reactions to being impacted by weapons, ragdolls causing a domino effect in enemies, and more combat upgrades are included here. Watch the trailer below to catch the carnage in action:

On top of that, a gear transmog system to switch the look of gear pieces without affecting their stats and a brand-new weapon crafting system involving blueprints and Craftmaster NPCs have arrived too. Lastly, bounties are incoming via a web portal, where players can complete various in-game objectives to earn unique weapons and outfits. Read the Gut Feeling changelog to read all the changes included in the massive update.

Since launch, one of the main areas the title has been receiving criticism from fans has been regarding the weight of combat, where zombies do not react satisfyingly to being hit or sliced. The long-awaited upgrades should remedy that issue quite well.

The Dying Light 2 Stay Human - Gut Feeling update is now available across PC and consoles. Techland will be hosting a special live stream discussing the update at 11 a.m. PDT on its Twitch channel today.

Dying Light 2 upgrades its physics and combat brutality in major update

Google employees repeatedly criticized the company’s chatbot Bard in internal messages, labeling the system “a pathological liar” and beseeching the company not to launch it.

That’s according to an eye-opening report from Bloomberg citing discussions with 18 current and former Google workers as well as screenshots of internal messages. In these internal discussions, one employee noted how Bard would frequently give users dangerous advice, whether on topics like how to land a plane or scuba diving. Another said, “Bard is worse than useless: please do not launch.” Bloomberg says the company even “overruled a risk evaluation” submitted by an internal safety team saying the system was not ready for general use. Google opened up early access to the “experimental” bot in March anyway.

Bloomberg’s report illustrates how Google has apparently sidelined ethical concerns in an effort to keep up with rivals like Microsoft and OpenAI. The company frequently touts its safety and ethics work in AI but has long been criticized for prioritizing business instead.

In late 2020 and early 2021, the company fired two researchers — Timnit Gebru and Margaret Mitchell — after they authored a research paper exposing flaws in the same AI language systems that underpin chatbots like Bard. Now, though, with these systems threatening Google’s search business model, the company seems even more focused on business over safety. As Bloomberg puts it, paraphrasing testimonials of current and former employees, “The trusted internet-search giant is providing low-quality information in a race to keep up with the competition, while giving less priority to its ethical commitments.”

Others at Google — and in the AI world more generally — would disagree. A common argument is that public testing is necessary to develop and safeguard these systems and that the known harm caused by chatbots is minimal. Yes, they produce toxic text and offer misleading information, but so do countless other sources on the web. (To which others respond, yes, but directing a user to a bad source of information is different from giving them that information directly with all the authority of an AI system.) Google’s rivals like Microsoft and OpenAI are also arguably just as compromised as Google. The only difference is they’re not leaders in the search business and have less to lose.

Brian Gabriel, a spokesperson for Google, told Bloomberg that AI ethics remained a top priority for the company. “We are continuing to invest in the teams that work on applying our AI Principles to our technology,” said Gabriel.

In our tests comparing Bard to Microsoft’s Bing chatbot and OpenAI’s ChatGPT, we found Google’s system to be consistently less useful and accurate than its rivals.

Source

Elon Musk threatens to sue Microsoft

Quantum effects of D-Wave’s hardware boost its performance

However, a new report from Bloomberg claims Google rushed out new AI products like Bard with little efforts to put in ethical guardrails for its information. The company was reportedly threatened with the sudden rise of ChatGPT, and the AI ethics team at Google was allegedly ignored when it decided to rush out Bard.

The article states:

The group working on ethics that Google pledged to fortify is now disempowered and demoralized, the current and former workers said. The staffers who are responsible for the safety and ethical implications of new products have been told not to get in the way or to try to kill any of the generative AI tools in development, they said.

The article says that before Bard was launched to the general public, Google employees were asked to test it. Some of their responses were very critical:

One worker’s conclusion: Bard was “a pathological liar,” according to screenshots of the internal discussion. Another called it “cringe-worthy.” One employee wrote that when they asked Bard suggestions for how to land a plane, it regularly gave advice that would lead to a crash; another said it gave answers on scuba diving “which would likely result in serious injury or death.”

In the end, Google launched Bard as an "experiment", telling users that it could make mistakes in its answers. However, it also shows that AI still has a long way to go before it offers answers that are truthful. A Google spokesperson responded to Bloomberg's request for a comment, stating, "We are continuing to invest in the teams that work on applying our AI Principles to our technology,”

New report claims Google is rushing out Bard and other AI products with poor ethical guards

The report from Bloomberg, citing unnamed sources, claims that millions of apps will be able to run on the mixed reality headset in its 3D interface, with just some minor tweaks that developers can make for those apps. That includes first party apps from Apple itself, including its Safari web browser, email, FaceTime, and even the Apple TV app. In fact, people will reportedly be able to watch movies and TV shows from the Apple TV app in a virtual location like a desert or even the sky while wearing the headset.

Apple is also reportedly working on a camera app that will let owners take pictures from the headset. It's also working on fitness and meditation apps, including having a virtual instructor on screen while you exercise. Gaming will also be a big app category for the device, The headset will also allow owners to run several apps on screen at the same time, according to the story.

The mixed reality headset is shaping up to be of Apple's biggest ever hardware launches. It may also be one of its riskiest launches as well, as the device is rumored to have a final price tag of $4,000 or even higher. Unconfirmed reports claim the headset will go on sale towards the end of 2023. However, sales of current VR headsets, like Sony's recently launched PS VR 2, are not that great, even at prices that are much lower than what Apple's devices is projected to be. It will be an uphill battle for Apple to convince consumers that this mixed reality headset a "must own" product.

Apple's mixed reality headset will reportedly support millions of current apps