Apple’s next custom hardware trick might be its own microLED screens

The terms Microsoft has come up with for the funding are very favourable to it indeed. Apparently, Microsoft would receive a massive 75% of OpenAI’s profits until it gets its original investment back. Once Microsoft recoups its initial investment, Semafor says “it would revert to a structure that reflects ownership of OpenAI”, that is, Microsoft would have a 49% stake and the investors would have another 49%, leaving the non-profit parent, OpenAI Inc. with just 2%.

Nothing is set in stone with this plan and it could completely fall through. It’s interesting that these details are coming to light, though, given that Microsoft was reportedly looking to integrate ChatGPT into Bing to give it an edge over competitors such as DuckDuckGo and Google. Neither Microsoft nor OpenAI commented on those rumours.

If Microsoft does go ahead with this $10 billion investment in OpenAI, it won’t be the first time it has helped the company financially. Back in 2019, Microsoft injected $1 billion into the company to help it build projects like ChatGPT.

Source

Adobe users have taken to social media to voice their concerns that the company is using the content they have saved on its servers to train its AI model, which the company calls Sensei. The model has been one of the Adobe suite’s key innovations of late, able to offer creative and impressive outputs. It now seems, however, rather predictably, that its skill and ability to impress Adobe users actually comes from the work of those very users, whose data has been training it.

The issue seems to have arisen when Adobe added a content analysis term to its data collection permissions, putting the impetus on users to opt-out of the plan. Hidden away in the terms and conditions, the company explains that it may analyze content using machine learning to develop and improve its products and services. This means that while Adobe claims that users maintain control over their privacy preferences and settings, in fact, it isn’t too easy for users to exert that control.

The move has clearly annoyed some users too with many listing ways to lock down privacy settings across Adobe's suite of Creative Cloud and Document Cloud apps in a bid to help other users fall foul of having their content used in this way.

This all raises the point of who owns what when it comes to AI, which is an issue that has been causing quite a debate among artists and creatives. If an AI has been trained on your work, surely your work has contributed to the creative output it has produced. In this sense then, surely you should have an element of ownership over it, shouldn’t you? Or is it the giant tech corporation that used your data to train its model that should reap all the rewards?

It is an interesting debate for sure, but it is slightly disheartening that as we move forward into implementations of seemingly revolutionary technologies, we are still seeing the same patterns of user data not really belonging to anybody and being fair game for tech giants to take and use at their will.

Source

Matter made a big splash at CES 2023.

Photo by Vjeran Pavic / The Verge

From Central Hall to The Venetian, Matter was the buzzword throughout CES 2023 this year, with most companies even remotely connected to the smart home loudly discussing their Matter plans (although a few were more subdued). The new smart home standard was featured in several keynotes and displayed prominently in smart home device makers’ booths as well as in Google, Amazon, and Samsung’s big, showy displays. 

 

More importantly, dozens of companies and manufacturers announced specific plans. Several companies said they would update entire product lines, while others announced new ones, sometimes with actual dates and prices. And Matter controllers have become a major thing, with at least four brand-new ones debuting at CES. Interestingly, nearly all of them have a dual or triple function, helping banish the specter of seemingly pointless white hubs stuck in your router closet.

 

Its undeniable momentum at the biggest consumer tech show of the year is one reason we named Matter The Verge’s “best in show” for CES 2023. And here, we’ve rounded up all the announcements from the show that, well, matter. 

 

In case you missed it, Matter is an open-source interoperability standard that allows smart home devices from any manufacturer to talk to other devices directly and locally with no need to use the cloud. This should make the smart home easier to set up, simpler to use, and more reliable to run. Matter works over the protocols Thread, Wi-Fi, and ethernet and has been jointly developed by Apple, Google, Samsung, Amazon, and pretty much every other smart home brand you can name, big or small. 

 

If a device supports Matter, it will work locally with Amazon Alexa, Samsung SmartThings, Apple Home, Google Home, and any other smart home platform that supports Matter. It will also be controllable by any of the four voice assistants.

 

However, Matter is still totally unproven, as there are very few devices anyone can actually get their hands on to test, so there is a lot of speculation still as to just how effective it will be. Plus, the initial Matter rollout since the launch in November has been complicated.

 

The big four have turned on Matter support on their platforms, but Amazon’s approach has been piecemeal, and aside from Apple, nobody supports onboarding devices to Matter on iOS yet.

 

However, that is shifting: at CES, Amazon announced a full rollout by spring, and Samsung’s Jaeyeon Jung told The Verge that Matter support is coming to its iOS app this month. There’s still no news on Matter support in Google Home’s iOS app. Then there’s the whole competing Thread network issue, although that sounds like it will be resolved sooner rather than later. 

 

Matter controllers — gadgets that will help people add Matter-compatible devices to their smart home — were popular products at CES.

Photo by Vjeran Pavic / The Verge

 

If all of these companies continue their support of Matter, then these early teething problems shouldn’t be more than that. After all, the rollout of a new wireless standard is never going to be easy. Just ask the Wi-Fi Alliance and Bluetooth Special Interest Group, which both sent representatives to the Connectivity Standards Alliance’s CES Matter party to show their support (and maybe sympathize). 

 

And the following cascade of announcements includes dozens of new products, so the Matter device drought should be over soon — although, judging by most of these ship dates, not until at least the second half of 2023.

Matter updates coming to existing products

• Amazon will update its Echo 4th-Gen smart speaker to be a Thread border router this spring and will expand its Matter support to more device types — including thermostats, blinds, and sensors in addition to light bulbs, plugs, and switches. All of its remaining compatible Eero and Echo devices and its iOS Alexa app will also become fully compatible with Matter in the spring. (Currently, you can control Matter devices with the Alexa app on iOS, but you can only onboard them using the Android app.)
• Leviton announced Matter support for its entire lineup of Wi-Fi light switches and plugs via a firmware upgrade later this year, starting with the Decora Smart Wi-Fi dimmer and switch. The remaining products will get their turn later this year.
• Lockly announced the Flex Touch Pro, a retrofit door lock with a built-in fingerprint sensor. Lee Zheng, CEO and founder of Lockly, told The Verge that Lockly’s entire line of fingerprint-reader Wi-Fi smart locks will receive an over-the-air update to support Matter later this year. That includes the Lockly Vision Elite and the Lockly Secure Pro.

 

The new Lockly Flex Touch Pro will get Matter compatibility with an over-the-air update, according to the company.

Photo by Jennifer Pattison Tuohy / The Verge

 

• Twinkly, of smart holiday lighting fame, announced its new entertainment mirroring app. The Entertainment Hub desktop app lets you sync visuals and audio to any Twinkly lights. Andrea Tellatin, CEO at Twinkly, told The Verge that it will make Matter updates available on all of its existing Wi-Fi smart lighting products and build Matter support into its new products. The next product to be released will be more budget-friendly RGB string lights called Twinkly Candies. These feature fun ball-, star-, and candle-shaped LEDs and are expected to launch in fall 2023. They will cost $49 per string and will ship without USB-C power adapters to keep prices down.
• Eve announced its Matter-enabled Thread products will arrive March 28th. The Eve Door and Window contact sensor, Eve Energy smart plug, and Eve Motion sensor will work with Matter out of the box. The company also announced that its Eve MotionBlinds will be upgraded to Matter in late Q1. Also on March 28th, Eve will launch the Eve MotionBlinds Upgrade Kit for Roller Blinds plus new smart honeycomb and Venetian blinds. Matter support will come to all the MotionBlinds products in Q1. The Venetian blinds are interesting because they’ll be the first smart slatted blinds that can raise and lower remotely rather than just tilt.
• Lock manufacturer Yale confirmed that its Thread module has received Matter certification and should be coming soon. You can sign up here to be notified. The module can be inserted into the company’s Yale Assure Lock and Yale Assure Lock 2 lines (except for the Lever Lock) and turn them into Matter-compatible locks. 

 

The new Yeelight Cubes are modular smart lights that will support Matter.

Photo by Jennifer Pattison Tuohy / The Verge

 

• European IoT device maker Shelly announced a slew of new products, including a new smart smoke alarm. Svetlin Todorov, CEO of Allterco Robotics US, which makes Shelly products, told The Verge that all of its Plus and Pro devices will support Matter with an optional firmware update that will roll out at the end of Q2. However, he cautioned that some features will not be available if the customer switches to Matter, including Shelly’s somewhat unique support for micro-JavaScript. This allows users to program customized functions directly on each device. He did say that the Matter update would be reversible.
• Smart lighting manufacturer Yeelight announced its new Cube Smart Lamp with Matter support as well as new products for its Yeelight Pro smart lighting series, which will support Matter with an over-the-air update before Q2 2023. Yeelight also announced an automatic curtain opener, which won’t work with Matter at launch.

New Matter devices announced at CES

Nanoleaf’s new Essentials range — inexpensive smart lighting including A19, lights strips, and BR30 bulbs — launches later this year with Matter support built in.

Photo by Jennifer Pattison Tuohy / The Verge

 

• Aqara — whose smart home hub will be updated to support Matter this month — unveiled a new Presence Sensor FP2 ($60, coming Q2), T1 LED light strip, and U100 smart door lock ($160 to $200, Q2). The company also showed off the G4 video doorbell ($120, February). Aqara said all of these devices will support Matter when their respective device types are added to the Matter spec. The company’s Thread-based P2 door / window contact sensor and P2 motion / light sensor will also work with Matter and were on display at CES and are slated for release in early 2023. No pricing has been announced.
• Nanoleaf announced its first Thread-based smart switches, which will be Matter compatible. The Sense Plus Smart Light Switch and Sense Plus Smart Wireless Light Switch will work with the new Nala Bridge, which is also a Thread border router and a motion sensor / night light. The company’s new Nanoleaf Skylight modular ceiling panel is also Matter compatible over Wi-Fi and can act as a Thread border router. Nanoleaf confirmed its Matter-enabled, Thread-powered Essentials light bulbs and light strips will arrive in Q1 and that it will be upgrading its existing modular light panels and light bars (Shapes, Elements, Canvas, and Lines) to Matter.

 

GE Lighting will start manufacturing its Cync A19 colour smart bulb and indoor smart plug with Matter support later this year.

Photo by Jennifer Pattison Tuohy / The Verge

 

• GE Lighting confirmed it will produce Matter-compatible A19 Cync Full colour Direct Connect Smart Bulbs and Cync Indoor Smart Plugs later this year, but it will not upgrade existing devices to support Matter. While it does plan to bring more Matter-compatible products to market this year, its new Dynamic Effects products, announced at CES and coming later this year, won’t have Matter support at launch. 

 

Govee is launching its first Matter product — the LED Strip Light M1.

Photo by Jennifer Pattison Tuohy / The Verge

 

This is because Matter doesn’t currently support the type of dynamic lighting effects that are the main feature of the lights. Patrick Miltner of GE Lighting told The Verge that the company might change that plan if and when Matter supports more advanced lighting functions. He also said it could upgrade all of the existing Direct Connect Cync lighting products to Matter with an over-the-air firmware update and will evaluate the benefit of doing that as Matter support rolls out more broadly across the industry.

 

• TP-Link announced it will release a Matter-enabled Tapo P125M Mini Smart Plug, Kasa KP125M smart plug, Tapo S505D Smart Dimmer Switch, Tapo P306 Smart Wi-Fi Outlet Extender, and the Tapo L535E Multicolour Smart Light Bulb in 2023. (Yeah, I still don’t get the difference between the Tapo and Kasa brands, either.)
• Smart lighting company Govee introduced its first Matter-compatible product, the Govee LED Strip Light M1. It also confirmed it will release new products that support Matter this year but will not upgrade its existing line of smart colour-changing lighting products. 
• Sengled announced it will release a new Matter version of its multicolour A19 smart bulb in Q1 of this year. Ted Zhang, vice president of Sengled, US market, told The Verge that the company’s current A19 bulbs will not be upgradable to Matter, but the company does plan to release more Matter-compatible products later this year. Pricing for the A19 Matter bulb will be approximately $14.99 for a one-pack and around $27.99 for a two-pack, he said.  

New Matter controllers that launched at CES

Matter controllers are required to onboard and control smart home devices in Matter. There are already a lot of Matter controllers out there — including 17 of Amazon’s Echo devices, Apple’s HomePod Mini and Apple TV line, and Google’s Nest Hub smart displays and Nest speakers. Some Matter controllers are also Thread border routers and can onboard Thread devices to Matter; others work over Wi-Fi only. 

 

• Samsung was all in on Matter at CES, with nearly its entire booth dedicated to its SmartThings platform and a large section featuring the launch of its $60 SmartThings Station. This is a SmartThings hub that supports Matter and Thread and doubles as a 15W fast wireless charging pad for Galaxy smartphones and earbuds. You can read more about it in our first look / early hands-on here.

 

The new SmartThings Station is a wireless fast charger and a SmartThings / Matter smart home hub in one.

Photo by Owen Grove / The Verge

 

• Samsung also confirmed its Family Hub fridge line will be updated to be a Matter-over-Wi-Fi controller this month. 2022 Samsung TVs and monitors will get the update in March, and the SmartThings Hub Dongle — which is designed to work with Samsung fridges and TVs to add Thread and Zigbee support — will be updated in the first quarter of this year. However, the TVs and fridges will not be Matter devices, so they won’t be controllable by other Matter apps and platforms.
• Starting this year, new Samsung products, including TVs, monitors, and smart fridges, will support Matter, Zigbee, and Thread natively. The company is building a SmartThings Zigbee and Matter / Thread One-Chip Module into all of its products, so you won’t need the separate SmartThings dongle.
• Other TV manufacturers announced or reconfirmed support for Matter as a controller in their 2023 and newer TVs, including LG through its new webOS 23 operating system and Hisense, which joined the CSA and said it plans to bring Matter support to products in 2023, including Hisense and Toshiba televisions.
• SwitchBot has a new Matter-enabled hub — the SwitchBot Hub 2 — coming in March for $69. The Wi-Fi-based hub will bridge SwitchBot Bluetooth devices to Matter, starting with the SwitchBot door lock, curtain controller, and SwitchBot Bot devices. In keeping with the trend for hubs to be multipurpose devices, the Hub 2 is also a temperature and humidity sensor with a large LED display.

 

TP-Link’s first smart home hub will support Matter and acts as a stand for a tablet so you can easily access smart home controls and other apps.

Photo by Jennifer Pattison Tuohy / The Verge

 

• The second-generation Mui Board, scheduled for release later this year, will support Matter as a controller over Wi-Fi for select devices, including lights. Preorders for the new Mui Board will begin this June through Kickstarter for $599. Devices should ship in November.
• The Homey Pro smart home hub is allegedly finally coming to the US in Q1, with support for Wi-Fi, BLE, Zigbee, Z-Wave, 433MHz, and infrared at launch. Matter and Thread support are scheduled for Q2 (Matter) and Q3 (Thread). Homey says the Homey Pro will act as a bridge to bring any non-Matter device into Matter (as long as it’s a device type supported by Matter). You can preorder it for $399 now, with shipping slated for February. 
• TP-Link showed off its first smart home hub — the Homebase Tapo H900 (no pricing or release date). The triangular orange device works as a Matter controller and a Thread border router. The hub also supports Wi-Fi and Bluetooth and is designed to control Tapo smart home devices. It doubles as a video storage device for Tapo cameras and also works as a stand / charger for a tablet, similar to Google’s upcoming charging speaker dock. This should allow you to access the Tapo app and control your smart home directly from the hub. 

What’s next for Matter?

 

A Nanoleaf logo next to a Matter logo. Companies’ continued support of the new standard will be crucial to its success.

Photo by Vjeran Pavic / The Verge

 

For Matter to work, it needs mass adoption in the tech industry, and not every company is jumping in with both feet. A few are dipping an early toe in, while others are putting their feet up and waiting to see what the standard can really do for them. A handful have had very little to say at all.

 

Also, some companies that suggested they could offer Matter updates for existing products have walked that back. That’s in part because the Matter spec changed and in part because of the complexity of sending a firmware update to customers’ devices that removes existing integrations and makes them have to re-pair it with all of their smart home services (something Eve had to deal with during its rollout of firmware updates last year). 

 

GE Lighting’s Miltner told The Verge at CES that the company doesn’t see the use in updating existing products to Matter as “it will break their existing functions.” Customers will have to reset devices to work with their favorite platforms. However, looking forward, Miltner sees the value. “It’s one specification for every single product; it’s the massive piece the smart home has been missing for so long.” Once Matter hits critical mass, he believes there will be plenty of benefits for GE Lighting and its customers.

 

A number of companies are taking this wait-and-see approach. “It benefits Schlage more to wait to see how things are going to come out first,” Paul Wilkie, a spokesperson from lock manufacturer Schlage, told The Verge. “That’s not to say that there’s not already something in development — it’s clear that there is something in development. But we are waiting to see how everything congeals over the next year.”

 

He pointed out that it’s one thing for smart lighting and plug manufacturers to jump on a new bandwagon, but for a device that secures your home like a lock, it’s important to make sure people are ready for the new tech.

 

While Matter wasn’t designed to leave existing smart home devices behind, it’s appearing more likely the most benefit will be for the future smart home rather than the past or present one. “We have a go-forward strategy for Matter,” Tobin Richardson, president of the CSA, which manages the Matter standard, confirmed to The Verge. “Moving forward, Matter solves a lot of compatibility issues, but in terms of backward compatibility, not every device will be upgradeable. You may just be doing it through a gateway.” He says that should be a straightforward process if manufacturers upgrade their bridges.

 

For those of you with robust smart home setups that you want to bring into Matter, bridging is going to be the most reliable route for backward compatibility. There aren’t a lot of solutions yet, but a few companies — including Aqara and Philips Hue — have announced their platforms will support bridging into Matter.

 

It’s also likely we’ll see dedicated bridges coming out that can bring Z-Wave and other products with proprietary protocols into Matter. Silicon Labs announced a new Unify SDK that will provide both a hardware and software solution for manufacturers to do exactly this.

 

One company for whom bridging would make a lot of sense is Lutron, with its popular Caseta smart lighting line of switches and plugs. Adam Mack of Lutron Electronics told The Verge that while the company is a member of the CSA and is watching the rollout carefully, it has no announcements around adding support for the standard to its product line. Which, to be fair, is one of the most interoperable and reliable smart home products on the market today.

 

Aqara’s new G4 video doorbell will support Matter, according to the company. However, cameras are not supported by Matter yet.

Photo by Jennifer Pattison Tuohy / The Verge

 

It’s also clear that Matter needs to step up its support for more functions and device types. Currently, only basic functions are supported. Lights, locks, sensors, etc., will have basic control capabilities: on / off; lock / unlock; motion / no motion; brighten / dim. But Matter doesn’t support advanced features such as dynamic lighting effects, adaptive lighting, shared access codes for door locks, and energy management for smart plugs.

 

Those who want to use those features will need to either choose a platform that supports them (for example, the Apple Home app supports adaptive lighting) or use the manufacturer’s app for that function (for example, the Eve app supports energy management for the Eve Energy plug). 

 

We are also waiting on new device types to be supported in Matter. Initially, there will be support for light bulbs and light switches, plugs and outlets, door locks, thermostats, blinds and shades, sensors (motion and contact), bridges, and TVs.

 

The CSA told The Verge that its next release, slated for spring, will include support for battery performance (so you can keep track of battery life or devices such as sensors), white goods home appliances (such as fridges, washing machines), and robot vacuums. But cameras and garage door controllers, which were previously announced as coming next, look like they’re still going to be a while.

 

There are also a number of categories in the initial Matter spec with few or no actual products announced, like thermostats, for example. There’s the Nest Thermostat (new) and a couple of radiator options that have been promised. But none of these have been updated yet and no timeline has been given. The Verge spoke with Ecobee CEO Stuart Lombard, who confirmed that its line of smart thermostats will support Matter but wouldn’t give any further details. Previously, the company has said its thermostats could be upgraded to Matter with a firmware update and could support Thread. Amazon hasn’t said if its smart thermostat will be upgraded.

 

So, what is the biggest challenge for Matter now that it’s had its big public debut and seemingly passed its big CES test? “The complexity can be a concern,” Richardson, of the CSA, told The Verge. “What we’re trying to do is straightforward. How IoT happens can sometimes be so complex and confusing that consumers back away.”

 

If every company continues to be invested in Matter and “doesn’t back into proprietary shells,” Richardson feels confident Matter will achieve all it set out to do. However, “the commercial teams are now coming in and looking for the differentiations,” he said. “I have to remind them all that while monetization is important, this is building a market, not a way to shut others out.”

 

Matter may not be the solution to the smart home’s challenges that every company wanted. But it is the solution they got. Now, we wait and see if it’s the solution they can actually deliver.

 

Why Matter mattered at CES

With all the recent excitement that has been gripping the internet, it’s not hard to see why Microsoft thinks incorporating ChatGPT into its suite of productivity apps is a good idea. The potential use cases are vast. For example, the model could be used to generate text in a similar way to how current tools like Jasper and Lex are already doing. This could include rewriting documents for clarity and even producing entire documents according to short text prompts input by the user.

However, there are other possibilities too. According to the reports, Microsoft looking to incorporate the chatbot into Outlook to "provide more useful search results when Outlook email customers look for information in their inboxes." Further possibilities could also include analyzing data in Excel or translating texts more reliably.

However, ChatGPT's rise to prominence has also caused concern among industry experts and educators due to its unreliable accuracy. OpenAI’s own CEO, Sam Altman has also warned against going in too hard on the tool and over-relying on the text it creates. While the chatbot can generate text that sounds plausible, it is not always correct. This is something that is likely to stand in the way of the technology being taken up being implemented in industries where companies need to rely on factually accurate information.

It is important to note here, however, that ChatGPT is an early version of the technology, and has been put out into the public domain specifically so that the developers can learn about these types of issues. This move from Microsoft sends a clear indication of its faith in this process and belief that this type of tech can revolutionize the day-to-day life of people who use the Office suite of programs and send emails every day.

As for a timeline for when we might see ChatGPT integrated into Office, there's no fixed estimate at this point. It's clear that the technology still has a way to go before it will be ready for mainstream use. Despite that, Microsoft’s move would be a big step toward making it more mainstream, which is why Office users should keep an eye on this development. The decision to incorporate ChatGPT into Office and Bing is a significant step toward making these kinds of generative AI tools more prevalent in our everyday lives.

Source

Microsoft’s new AI can simulate anyone’s voice with 3 seconds of audio

Raspberry Pi upgrades its Camera Module with HDR, autofocus, and more

Policymakers in Taiwan have passed laws to allow chip companies take 25% of  yearly research and development costs and turn them into tax credits, Bloomberg reports. This is part of an attempt to make sure that cutting-edge chip tech stays in Taiwan, as other nations have pushed incentives to strengthen their own supply chains.

"The Ministry of Economic Affairs stated that Taiwan is an important link in the global supply chain and a long-term reliable partner of international manufacturers, which is unique and irreplaceable," the Ministry of Affairs wrote in a release. It says that the move comes "In the face of huge incentive measures proposed by the United States, Japan, South Korea, and the European Union."

Chipmakers in Taiwan will also be able to file for tax credits on 5% of their annual costs for purchasing equipment for advanced nodes. (This equipment makes up some of the costliest parts of upgrading fabs or building new ones.) Bloomberg notes that credits earned can't surpass 50% of a company's annual income taxes.

The measures were introduced late last year and are likely to take effect this year.

The United States passed the CHIPS and Science act in July to support domestic chip production. Since then, plans for a number of American fabs have popped up, including Intel spending over $40 billion on facilities in the U.S. in Arizona, Ohio, and New Mexico. Taiwan Semiconductor (TSMC) announced a 5 nanometer-capable fab near Phoenix in mid-2020, which could benefit. GlobalFoundries, Samsung and Texas Instruments are also set to upgrade or set up new plants in the United States. The Semiconductor Industry Association, a lobbying group for the industry, says the CHIPS act attracted $200 billion in private investments to the sector.

Abroad, Intel had planned for a "mega-fab" in Magdeburg, Germany, receiving roughly $7.3 billion in funds from the European Chips act. In December, Intel reportedly backed away from starting construction due to a "difficult market situation."

While having chips made domestically has long been considered important to national security, the world saw what happens when supply chains back up in the early phases of the Covid-19 pandemic, which created a shortage of chips for computers, cars and other goods. That, mixed with a dash of geopolitical tension, has had companies looking to diversify their supply chains, and it looks like several of them will be able to follow the money around the world.

Source

Today, in a bid to showcase its commitment towards innovative, modern datacenters, the tech giant has acquired Fungible - a California-based firm that specializes in the development of various forms of hardware and software that are meant to improve the efficiency of datacenters.

In a press release, the team at Fungible expressed hope in association with the move, noting:

"We are proud to be part of a company that shares Fungible's vision and will leverage the Fungible DPU and software to enhance its storage and networking offerings. We would like to thank our loyal employees for their dedication and hard work over these last seven years and our customers, partners, and investors for their belief and support in our technology."

Fungible developed its data processing unit (DPU) in 2016, focusing on improving computational efficiency for data-centric operations within server nodes in scale-out datacenters. Microsoft believes that improvements that have been made throughout the DPU manufacturer's journey will enable the Fungible team - together with the Redmond company's own infrastructure engineering team - to innovate in datacenter infrastructure, ensuring that network and storage performance can be enhanced.

Security, reliability, and scalability will remain core focuses for Microsoft as well, as it looks to add to its portfolio of datacenter innovation-centric investments for the long term.

Source

Novelist Ning Ken first saw Beijing’s Zhongguancun neighborhood in 1973 as a 14-year-old on a school trip to the Summer Palace, former imperial gardens looted by European troops during the Opium Wars. “At that time, once you passed the zoo, Beijing was just countryside and farmland,” he says, recalling the bus ride heading northwest. Out the window and amid the fields, Ning saw the campuses of China’s most prestigious research institutions, which had birthed China’s nuclear program and hydroelectric dams. They included the Chinese Academy of Sciences and Peking and Tsinghua Universities. 

Today that stretch of road is the heart of China’s technology industry, a busy neighborhood with a subway stop and glass towers housing Chinese and Western tech companies. The neighborhood’s transformation mirrors the dramatic changes to China’s economy and culture over the past four decades. Tech companies that grew out of Zhongguancun expanded the boundaries of how businesses could operate—often by staying one step ahead of regulators—and came to shape Chinese power overseas.

In the West, coverage of China’s tech industry often focuses on how it is restricted or controlled by the government. In Ning’s telling, the innovators of Zhongguancun helped “liberate” the Chinese people from the strictures of a fully state-run economy by carving out a path for entrepreneurship as the country tentatively opened up.

When the first tech companies were established in Zhongguancun in early 1980s, every industry was state-owned, and every aspect of a person’s life was dictated by their danwei, or work unit, from where they lived to whom they married. When an entrepreneur named Wang Hongde left his research position at the Chinese Academy of Sciences in 1982 to start an IT company, taking several colleagues with him, it “tore a crack in the old system,” Ning says.

Two generations later, Zhongguancun and the rest of China are almost unrecognizable. People can chase fortunes and change careers in ways that would have been unthinkable in the early 1980s. Recent events have shown that change can still happen fast, with pressure from the bottom up, enabled in part by some Zhongguancun social media companies. In late November, people in cities across the country staged protests against extreme zero-Covid measures. Restrictions that after three pandemic years seemed permanent soon toppled, and China began reopening.

Red-Light Revolution

Ning, a native Beijinger, has published several acclaimed novels in China, but his first book to be translated into English is Zhong Guan Village: Tales From the Heart of China’s Silicon Valley, a nonfiction account of Zhongguancun’s history. It introduces the entrepreneurs and academics who built China’s tech industry, from the early days of Deng Xiaoping’s reform and opening up policies in the late 1970s to more recent boom times, when Chinese tech firms like search giant Baidu and TikTok’s parent company ByteDance grew out of the neighborhood. 

Many of the people Ning introduces aren’t household names outside of China, but their stories illustrate how Zhongguancun’s entrepreneurs found clever ways to work within and around the system. Today, many are celebrated for their role in opening China’s economy and advancing its technology industry. “I want this book to not only show the path of reform and opening up over the past 40 years but also to show readers the spiritual wealth of these individuals,” he says, writing to WIRED in Chinese. “I’m a novelist. The core of my interests is always people, predicaments, growth, emotions, psychology, and the way society and history relate to those things.”

One of those individuals is nuclear physics professor Chen Chunxian, widely credited as the father of today’s Zhongguancun. In 1978, Chen traveled to the US and visited Silicon Valley in California and Route 128, the tech corridor near Boston that grew out of MIT and Harvard. He came back inspired to build China’s own university-adjacent tech zone along the route Ning took to the Summer Palace, which has since been renamed Zhongguancun Street.

Private businesses weren’t strictly legal at the time, but in 1980 Chen devised a workaround. He started a “service department” inside the academic Beijing Plasma Physics Association, providing a way for professors to work outside of their regular hours providing consulting and IT services, and later manufacturing electronic components. 

“You can’t drive forward a revolution without going through a few red lights; revolution is all about breaking the old rules,” Ning quotes Zhao Qiqui, the lead manager of the service department, telling Chen. The department drew government scrutiny for profiteering off public academic research—some professors resigned and Chen feared prosecution—but was eventually hailed by officials as a model for innovation in science and technology.

That cycle of tech entrepreneurs pushing legal boundaries and authorities later embracing and then regulating their new way of doing business has regularly repeated itself in China ever since. Lenovo was founded in a bungalow that was formerly a janitor’s office at the Chinese Academy of Sciences and, like other early Zhonguancun tech companies, operated in a legal gray area. Each existed under the umbrella of a state-owned company but operated like a private business, until 1990s reforms made fully private enterprises legal. More recently, in 2004, Jack Ma, founder of ecommerce giant Alibaba, launched a digital payments system called Alipay that was flat-out illegal in China, reassuring employees that if anyone went to prison, it would be him. Ma did not go to jail, and Alipay helped Chinese ecommerce to flourish, outpacing US tech companies in developing a widely used mobile payment system.

Exceeding Fantasy

Ning’s own career has been shaped by the transformative power of technology. In the book he describes signing up for Yinghaiwei, China’s first ISP, in 1995. “I remember it very clearly,” he writes, “how I dialed the number and heard the swift rhythm of the connecting tone given off by the connection icon on the Windows 95 interface display.” Ning joined an online literature chatroom and began uploading his work. In 2000, after literary magazines rejected his novel City of Masks, it was serialized on Sina, a successor to Yinghaiwei, garnering half a million hits in a month.

That book—about a high school graduate obsessed with Alfred Hitchcock and Sherlock Holmes who rambles across China trying to unravel his own family mysteries—was eventually published in hardcover, and Ning became known for his experimental, ambitious prose. “He writes for himself,” says Thomas Moran, a professor at Middlebury College who is translating Ning’s novel The Tibetan Sky into English. “He doesn’t seem interested in writing for any assumed audience, whether inside China or in translation. He doesn’t seem to care what the critics think or what the censors might think.”

With the turn of the millennium, as Ning's literary career began to take off, Chinese tech companies began to operate more like their Western counterparts. Lenovo, by then far bigger than a bungalow, exemplified the change when it acquired IBM’s personal computer business in 2005. Even acquiring a small portion of IBM’s business seemed, Ning says, “like a snake swallowing an elephant.” When the acquisition proved successful, Lenovo founder Liu Chaunzi was seen as a national hero.

Soon after, a new generation of founders came to the fore who benefitted from three decades of opening and growth. In 2012, an ambitious employee at ecommerce company Alibaba was inspired to start the ride-hail company Didi after he arrived drenched for a business meeting in Hangzhou because he couldn’t find a cab. The service launched before receiving approval from Beijing’s traffic regulator and hired people to spend their days hailing cars to convince drivers that signing up would pay off. 

Zhong Guan Village was originally published in China in 2017, when Chinese tech companies seemed unstoppable. It was two years after Ning had given a speech in which he coined the term “ultra-unreal”—in Chinese, 超幻 (chaohuan), which literally translates to beyond or exceeding fantasy—to describe the present-day China depicted in his surrealist novel Three Trios. In it a prison librarian recounts the stories of two inmates on death row: a corrupt CEO and the personal secretary to a provincial governor who is suspected of wrongdoing. Ning’s neologism captures how real news stories of rapid growth and extreme corruption in present-day China can seem so far-fetched as to border on the surreal.

Garage Startups

The English translation of Zhong Guan Village by James Trapp arrived last year, after the Chinese Communist Party appeared to curb the power of its tech giants. New regulations addressed some of their worst practices, such as sloppy data collection or blocking competitors from the “walled gardens” of their tech ecosystems. Didi, despite buying up Uber’s China operation and growing fast overseas, became a cautionary tale when the government levied massive fines for skirting competition laws and mishandling customer data. The company was forced to delist from the New York Stock Exchange less than a year after going public in 2021. 

That government turn against tech was echoed by some sections of the Chinese public. The founders of Lenovo and Alibaba began to lose their hero status. “Some people even think they are criminals,” Ning says. Lenovo was attacked by nationalist commentators for overpaying its executives and selling state assets at a loss.

Meanwhile, in the West, and particularly the US, impressions of Chinese tech companies tend to focus on theft of intellectual property or fears they are acting as arms of the Communist Party. There are documented cases of government-affiliated hackers stealing trade secrets to benefit Chinese industry, but Zhonguancun tech firms are too often unfairly written off as inferior copies of their Western counterparts. Less discussed are the ways they have made the Chinese economy and culture more open—or created good products. Despite the often vague national security concerns voiced about TikTok by US politicians, ByteDance cofounder Zhang Yiming appears to be foremost a tech product visionary. He has created an app so effective that more than a billion people are hooked, and executives at Meta and other Silicon Valley companies are terrified their products are becoming irrelevant.

Late in the book, Ning visits Garage Cafe, a Zhongguancun institution with exposed pipework where people who might become the next Zhang Yiming go to chase startup dreams. Ning describes “computers everywhere the eye could see in 800 square meters of workspace, almost half a football field, with entrepreneurs and investors thinking up ways of working together, cell phones pressed to ears.” Alongside coffee, Garage offers workshops, a library of programming books, rental of smartphones and other devices by the hour—even its own startup incubator. “Garage Cafe left a vibrant impression on me,” he says. “The older generation broke free of their chains to walk the earth. The young people at Garage Cafe are flying.”

Ning concedes that the years since his book was written have been hard on China’s young innovators, who have had their lives and ambitions restricted by zero-Covid measures. But in recent weeks, the streets of Beijing and other cities have come to life again. In the interim, the government has imposed new restrictions on recommendation algorithms, AI-generated content, the gaming industry, and online fan clubs. But looking back at Zhongguancun’s history suggests that those restrictions may only inspire tech entrepreneurs to get more creative.

What the West Doesn’t Know About China’s Silicon Valley

(May require free registration to view)

Dell's 6K USB-C monitor was the one of the most tantalizing displays at CES 2023.

Scharon Harding

 

The Consumer Electronics Show (CES) never fails to deliver a pile of new gadgets and gizmos, but finding products that bring something new and valuable to the table can be a real challenge. CES 2023 had its share of product refreshes, clones, and minor updates, but this year also proved there's still some "wow" factor to be found at the tech show.

 

And that includes the event's PC monitor selection. All the monitors on this list are promised to be real products coming out this year. Better yet, they all have some unique features that aren't readily available to consumers today.

 

Here are the seven most intriguing monitors from CES 2023.

Asus ProArt Display OLED PA32DCM

I ended 2022 begging for a better OLED monitor selection in 2023. Those pleas have already started to be answered.

 

There haven't been too many OLED monitors smaller than 40 inches. That has meant limited OLED options for people looking for something to put on their desks for productivity and creative work. CES 2023 brought several desktop-sized OLED monitors, but most target PC gamers. That includes Acer's Predator X27U (27 inches), Asus' ROG Swift OLED PG27AQDM (27 inches), and MSI's MEG 342c QD-OLED (34-inch ultrawide).

 

Asus' ProArt Display OLED PA32DCM is a 31.5-inch 4K monitor aimed at creative professionals. Asus' blogmarkets the panel toward photographers, filmmakers, and people working with HDR content, comparing it to a reference display. It attempts to earn that not-so-humble brag with a claimed colour error of Delta E <1 (many premium monitors claim a Delta E of <2 or 3). Asus also says the display has 99 percent DCI-P3 colour coverage and a max brightness of 700 nits.

 

The company touts the OLED screen's use of an RGB stripe panel rather than RGBW, which can hurt text clarity, particularly in RGB-optimized Windows or the much-maligned PenTile layout. The tech is also different from that seen in the QD-OLEDmonitors that started releasing last year, largely targeting gamers seeking vibrant colours. It's exciting to see a new productivity-focused OLED option besides LG's UltraFine OLED monitors hit the market.

 

Asus said it's looking to get the PA32DCM VESA DisplayHDR True Black 500 certification, which guarantees 500 nits in a 10 percent pattern and 300 nits full-screen with an ultra-low black level. That would help set it apart from LG's OLED.

 

Some monitors announced at CES 2023 feature 140 W USB-C Power Delivery (PD), so the PA32DCM's 90 W Thunderbolt 4 port isn't very impressive. It's still sufficient for powering many ultralight PCs, though. We don't know the full port selection yet, but there should be a 15 W Thunderbolt 4 port and HDMI.

 

Asus didn't reveal the monitor's price but said it would come out in the first quarter of the year.

The 7 most interesting PC monitors from CES 2023

England just made gigabit internet a legal requirement for new homes

Our brain has evolved to make predictions and explanations in unstable and ill-defined situations. For instance, to understand a novel situation, the brain generates a single explanation on the fly. If this explanation is upturned by additional information, a second explanation is generated. 

Machine learning, on the other hand, typically takes a different path: It sees reasoning as a categorization task with a fixed set of predetermined labels. It views the world as a fixed space of possibilities, enumerating and weighing them all. This approach, of course, has achieved notable successes when applied to stable and well-defined situations such as chess or computer games. When such conditions are absent, however, machines struggle.

One such example is virus epidemics. In 2008, Google launched Flu Trends, a web service that aimed to predict flu-related doctor visits using big data. The project, however, failed to predict the 2009 swine flu pandemic. After several unsuccessful tweaks to its algorithm, Google finally shuttered the project in 2015.

In such unstable situations, the human brain behaves differently. Sometimes, it simply forgets. Instead of getting bogged down by irrelevant data, it relies solely on the most recent information. This is a feature called intelligent forgetting. Adopting this approach, an algorithm that relied on a single data point—predicting that next week’s flu-related doctor visits are the same as in the most recent week, for instance—would have reduced Google Flu Trends' prediction error by half. 

Intelligent forgetting is just one dimension of psychological AI, an approach to machine intelligence that also incorporates other features of human intelligence such as causal reasoning, intuitive psychology, and physics. In 2023, this approach to AI will finally be recognized as fundamental for solving ill-defined problems. Exploring these marvelous features of the evolved human brain will finally allow us to make machine learning smart. Indeed, researchers at the Max Planck Institute, Microsoft, Stanford University, and the University of Southampton are already integrating psychology into algorithms to achieve better predictions of human behavior, from recidivism to consumer purchases. 

One feature of psychological AI is that it is explainable. Until recently, researchers assumed that the more transparent an AI system was, the less accurate its predictions were. This mirrored the widespread but incorrect belief that complex problems always need complex solutions. In 2023, this idea will be laid to rest. As the case of flu predictions illustrates, robust and simple psychological algorithms can often give more accurate predictions than complex algorithms. Psychological AI opens up a new vision for explainable AI: Instead of trying to explain opaque complex systems, we can check first if psychological AI offers a transparent and equally accurate solution.

In 2023, deep learning in itself will come to be seen as a cul-de-sac. Without the help of human psychology, it will become clearer that the application of this type of machine learning to unstable situations eventually runs up against insurmountable limitations. We will finally recognize that more computing power makes machines faster, not smarter. One such high-profile example is self-driving cars. The vision of building the so-called level-5 cars—fully automated vehicles capable of driving safely under any conditions without human backup—has already hit such a limitation. Indeed, I predict that in 2023, Elon Musk will retract his assertion that this category of self-driving cars is just around the corner. Instead, he will refocus his business on creating the much more viable (and interesting) level-4 cars, which are able to drive fully autonomously, without human help, only in restricted areas such as motorways or cities specifically designed for self-driving vehicles. Widespread adoption of level-4 cars will instead spur us to redesign our cities, making them more stable and predictable, and barring potential distractions for human drivers, cyclists, and pedestrians. If a problem is too difficult for a machine, it is we who will have to adapt to its limited abilities.

It’s Time to Teach AI How to Be Forgetful

(May require free registration to view)

Apple may finally debut its mixed reality headset this spring

New year, new builds

After the holiday break spread over the last few weeks, Microsoft resumed rolling out Insider Preview builds a couple of days ago. Windows 11 Dev Channel Insiders were treated to build 25272, which contains a long list of bug fixes and general improvements. But more importantly, it also rolls back a controversial Start menu feature, that being the recommendation of websites in bottom half of the UI. That said, it's not clear yet if this will make a return in the future. Additionally, we learned that Microsoft may also be working on a "Gallery" interface, but it's unclear how it will differ from the Pictures directory.

Meanwhile, Beta Channel Insiders received build 22623.1095 (KB5022364). It contains ever more rounded corners for the search box in the Start menu, along with bug fixes for the Task Manager, Start menu, and the Taskbar.

That's not to say that there isn't anything to talk about in the generally available version of Windows. Microsoft fixed a SQL Server issue affecting apps on various versions of Windows in the past week too.

And for all you performance stats nerds out there, some new data indicates that Windows 11 may not be faster than Windows 10, and it may not have a huge advantage over Ubuntu either. In related news, there have also been sporadic reports about the latest Patch Tuesday update for Windows 11 version causing freezing issues on AMD Ryzen PCs. AMD has also claimed that it will boost Windows 11 performance on at least its latest Ryzen 7000X3D chips.

Lastly, if you're still using Windows 8.1 for some reason, consider upgrading as soon as possible as the OS is running out of support in a couple days.

Microsoft 365 improvements

There were a bunch of Microsoft 365 related updates in the past few days. For starters, the Microsoft 365 Current Channel netted build 15928.20198 (version 2212). The update brings new features for Access and Word and also fixes a few bugs in the latter software and Excel. Speaking of Excel, Microsoft has finally introduced the Automate tab for Excel desktop users both on Mac and Windows too, find out what it does here.

Meanwhile, Outlook mobile users are set to receive a bunch of new features next month. These include better email forwarding and better warning messages and cleanup options when you're about to run out of email storage. On the other hand, iOS users will receive more customization options.

Similarly, Team customers were treated to some good news too. The meeting toolbar has been revamped to make navigation a lot easier, but the feature is not generally available yet. Furthermore, February will bring "busy options", which basically means that users can customize how incoming calls are routed when you are busy in an existing call or meeting.

Finally, it's worth highlighting that Skype and OneDrive briefly went down this week but Microsoft managed to resolve the outage within a few hours.

Microsoft Edge and other app updates

With 2023 now underway, it seems like Microsoft has its work well cut out in the Edge department. This is because the latest stats from Statcounter have revealed that the browser managed to gain only 2% desktop market share throughout the whole of 2022. This does not mean that it performed worse compared to other browsers, but it does solidify the fact that having more than one billion Windows devices running worldwide is not enough to make Microsoft win this race.

In related news, Microsoft Edge received a new Dev build recently too. On Android, a notable feature is improved ad blocking but you should experience fewer crashes and enjoy other bug fixes across other platforms too.

Similarly, PowerToys 0.66.0 towed with it a hefty list of bug fixes and improvements - but no new toys unfortunately. And third-party utility Virtual Desktop Helper now supports Windows 11 Insider builds too. Find out more details about this update here.

Git gud

While Edge in the desktop market probably isn't doing as well as Microsoft would have hoped, Windows 11 is seemingly making waves in the gaming community. Valve's latest survey data indicates that the OS is being used by almost 29% of gamers, which is a significant figure given the young age of the operating system.

In other notable news, Microsoft has agreed to ZeniMax Studios staff forming a labor union in the U.S. This is a first for Microsoft's U.S. division, with the labor union consisting of approximately 300 quality assurance team members, more than any other U.S. game studio to date.

On the Xbox side, we got an Alpha ring preview build with "Recent Parties" feature to quickly dive back into a game with the same group of people you were playing with before. Meanwhile, CES 2023 saw a gimmicky ASUS ROG Raikiri Pro Xbox controller sporting a 1.3-inch OLED display above the Xbox button which can be used to display text or animated wallpapers.

There was also a new Xbox Series X bundle with Forza Horizon 5 and lots of in-game extras. Additionally, Games with Gold subscribers were treated to Iris Fall, but if console gaming doesn't really suit your palette, check out this Weekend's PC Game Deals curated by our News Editor Pulasthi Ariyasinghe instead.

Dev Channel

• Nvidia has announced new RTX Video Super Resolution for Microsoft Edge and Google Chrome
• A student is working on a tool to combat AI-powered plagiarism from tools like ChatGPT

• Microsoft and ISRO are collaborating to support space-tech startups in India

   

Under the spotlight

In this section, we first have a couple of Outlook-focused guides. The first is from News Reporter Hemant Saxena who describes how to recall or edit an email after sending it through the Outlook 365 app.

Next, we have a guide from News Editor Justin Luna talking about how to schedule emails from the Outlook web app.

Our third and final guide for this week is from forum member Adam Bottjen's Tech Tip Tuesday article, explaining how you can reduce the blue light from your PC's display to ease the strain on your eyes and help you sleep better. Read this important guide here.

Logging off

Our most interesting news item of this week is undoubtedly a report claiming that Microsoft is looking to integrate the popular ChatGPT large language model in its Bing search engine. This wouldn't be too surprising considering that Microsoft has invested $1 billion in OpenAI, the company behind the machine learning model. Of course, both companies have declined to comment on the topic but it would be really interesting to see if Bing on steroids powered by ChatGPT is enough to dethrone Google in the long term.

Microsoft Weekly: New builds for Windows 11, Microsoft 365 improvements, and Edge upgrades

Creative founder Sim Wong Hoo, the man behind Sound Blaster, has died

Data you’ll be able to send through satellites won’t include the full breadth of data that you’re used to sending on general mobile networks, instead, Snapdragon Satellite will enable emergency messaging SMS messaging, and other messaging applications. You shouldn't expect this on your mid-range phone either, Qualcomm says it’s coming to premium devices first, in select regions.

“Robust and reliable connectivity is at the heart of premium experiences. Snapdragon Satellite showcases our history of leadership in enabling global satellite communications and our ability to bring superior innovations to mobile devices at scale,” said Durga Malladi, senior vice president and general manager, cellular modems and infrastructure, Qualcomm Technologies, Inc. “Kicking off in premium smartphones later this year, this new addition to our Snapdragon platform strongly positions us to enable satellite communication capabilities and service offerings across multiple device categories.”

While the hardware will initially come to high-end phones, Qualcomm has expressed that Snapdragon Satellite could expand to other devices including laptops, tablets, vehicles, and IoT. The company said that OEMs and app developers will eventually be able to launch branded services that use satellite connectivity, but again, that will come further down the road by the sounds of it. Furthermore, Qualcomm says that Snapdragon Satellite will support 5G Non-Terrestrial Networks (NTN) as constellations become more widely available.

Source

Now, with the arrival of platforms like ChatGPT by OpenAI, which has been trained to generate long-form answers, many are using it to spin well-referenced and uncannily natural essays. Although OpenAI said that it would watermark ChatGPT output,

people like Edward Tian are determined to solve this issue of AI plagiarism for good.

GPTZero makes use of "perplexity" and "burstiness" to determine a bot-written text. Perplexity, in machine learning, refers to how surprised a language model is when it sees new data while the phenomenon that the use of a term once in a document is likely to happen again is called burstiness. Human writing has more burstiness leading to more sentence variation, and language models can predict text better when the perplexity is lower.

As pointed out by Tian in a Substack newsletter, over 10,000 people have tested out the publicly available version of GPTZero on Streamlit to date. He also mentioned that he had updated the GPTZero model to reduce false positives. Furthermore, he is working on further updates to GPTZero including new capabilities and scalability.

While Microsoft has plans for using ChatGPT's capabilities to its advantage, the New York City (NYC) education department has blocked access to ChatGPT on school devices and networks due to concerns regarding plagiarism.

Source

The technology can instantly determine whether a person has exceeded the legal alcohol limit purely on using a 12-seconds recording of their voice.

In a paper published in the journal Alcohol, the study led by Ph.D. student Abraham Albert Bonela and supervised by Professors Emmanuel Kuntsche and Associate Professor Zhen He, from the Center for Alcohol Policy Research and the Department of Computer Science and Information Technology at La Trobe University, respectively, describes the development of the Audio-based Deep Learning Algorithm to Identify Alcohol Inebriation (ADLAIA) that can determine an individual's intoxication status based on a 12-second recording of their speech.

According to Albert Bonela, acute alcohol intoxication impairs cognitive and psychomotor abilities leading to various public health hazards such as road traffic accidents and alcohol-related violence.

"Intoxicated individuals are usually identified by measuring their blood alcohol concentration (BAC) using breathalyzers that are expensive and labor-intensive," Albert Bonela said.

"A test that could simply rely on someone speaking into a microphone would be a game changer."

The algorithm was developed, and tested against, using a database dataset of 12,360 audio clips of inebriated and sober speakers. According to the researchers, ADLAIA was able to identify inebriated speakers—with BAC of 0.05% or higher—with an accuracy of almost 70%. The algorithm had a higher performance of almost 76%, in identifying intoxicated speakers with a BAC of higher than 0.12%.

The researchers suggest that one potential future application of ADLAIA could be the integration into mobile applications and to be used in environments (such as bars and sports stadiums) to get instantaneous results about inebriation status of individuals.

"Being able to identify intoxicated individuals solely based on their speech would be a much cheaper alternative to current systems where breath-based alcohol testing in these places is expensive and often unreliable," Albert Bonela said.

"Upon further improvement in its overall performance, ADLAIA could be integrated into mobile applications and used as a preliminary tool for identifying alcohol- inebriated individuals."

More information: Abraham Albert Bonela et al, Audio-based Deep Learning Algorithm to Identify Alcohol Inebriation (ADLAIA), Alcohol (2022). DOI: 10.1016/j.alcohol.2022.12.002

Source

AMD decided to have an important public keynote address at the start of the new year. While AMD announced a lot of things in. Laptop CPUs, GPUs, AI, embed among other things. But the biggest thing which excited us and what everyone was waiting for was finally announced by AMD.

In the keynote (YouTube video), AMD announced new three new 3D V-Cache based Ryzen 7000 series processors.

Ryzen 7950X3D, 7900X3D, 7800X3D Announced

AMD finally revealed the successor to its groundbreaking and game-changing Ryzen 5800X3D processor. Back in September, it was rumored that AMD was going to release not one, but three processors based on the 3D V-Cache technology, looks like those rumors were true.

Those unaware, 3D V-Cache technology is a tech by AMD where it stacks additional L3 cache on the top of the CPU. This gives the CPUs an extraordinary performance boost in some tasks like gaming. So predictably, their Ryzen 7000 versions are going to be exciting.

AMD Ryzen 7 7800X3D

AMD started the announcement by showing the Ryzen 7 5800X3D’s direct successor, the Ryzen 7 7800X3D processor.

The Ryzen 7 7800X3D is an 8 cores, 16 threads processor. It comes with a boost clock of up to 5GHz speeds. The boost speeds here is 500MHz higher than the Ryzen 7 5800X3D. AMD is not revealing the base clock of the Ryzen 7 7800X3D yet.

It comes with the same 96MB L3 cache found in its predecessor. It’s likely that it’s using the same 32MB base cache and an additional 64MB stacked 3D cache found in the Ryzen 5800X3D. The L2 cache too has been doubled from 4MB to 8MB.

The TDP, however, has been increased from 105W to 120W.

In gaming performance, AMD claims that when compared to the Ryzen 5800X3D, the Ryzen 7800X3D is on average 15% faster across popular games.

AMD Ryzen 9 7950X3D

The next on the list is the processor is the one which can potentially become the best gaming CPU in the world. AMD revealed the AMD Ryzen 9 7950X3D.

The AMD Ryzen 9 7950X3D is a 16 cores, 32 threads processor. It comes with a boost clock of 5.7GHz, which is the same as that of Ryzen 9 7950X. However, the base clock is 300MHz lesser in the 7950X3D.

The Ryzen 9 7950X3D comes with 128MB L3 cache, which is double than the 64MB L3 cache found in the Ryzen 9 7900X CPU.

Compared to the 7900X, the TDP of 7950X3D has been lowered from 170W to 120W. While there’s no mention, this basically means overclocking will be limited.

Interestingly, there’s something we found out by looking at the specs (you can credit us for it, thank you) is that AMD claims it’s not made for consumer use. Now, whether this is a mistake or an intentional one, we don’t know.

When it comes to gaming performance at 1080p, AMD claims that the Ryzen 7950X3D is up to 24% faster than the Intel Core i9-13900K CPU, which is the fastest CPU in the market currently. If these claims are true, then AMD surely is almost running away with the prize for the best gaming CPU. However, AMD has not spoken about average performance here. Also, the unlocked Intel Core i9-13900KS release is just around the corner.

AMD Ryzen 9 7900X3D

AMD also announced the AMD Ryzen 9 7950X3D processor. Though no specific slides were shown for it.

The AMD Ryzen 9 7900X3D processor is a 12 cores, 24 threads CPU. It comes with a boost clock of 5.6GHz, which is the same as the Ryzen 9 7900X. However, here too, the base clock has been lowered by 300MHz to 4.4GHz.

In L3 cache, it’s going to be the same as the Ryzen 9 7950X3D. The Ryzen 9 7900X3D features 128MB of L3 cache, which doubles the 64MB found in the Ryzen 9 7900X,

Here too, the TDP has been lowered from 170W to 120W.

Additionally, just like Ryzen 9 7950X3D, here too AMD claims that the Ryzen 9 7900X3D is not for consumer use.

Release Dates And Other Announcements

The AMD Ryzen 7950X3D, 7900X3D, 7800X3D processors are going to release this February. AMD has shared no details about the prices of these new processors. Pricing is something which will decide how they stack-up against their competitor Intel’s processors.

Additionally, AMD also announced new low powered 65W processors. The non-X Ryzen 7600, 7700 & 7900 CPUs. It adds up with new entry level AM5 motherboards. Which is really welcome as current Ryzen 7000 supporting motherboards are really expensive currently.

The 3D V-Cache Chiplet Confusion

The new announced processors are not without their concerns, though. The new AMD Ryzen processors follow a chiplet design. Where small chips are joined together to make a CPU.

The biggest a chiplet can get is 8 cores in the Ryzen 7000 series. It means that while the Ryzen 7 7800X3D comes with a single chiplet, both Ryzen 9 7900X and 7950X come with two 8 core and 6 core chiplets respectively. So the question is, how will AMD stack up the new additional 3D V-Cache on them.

It has now been revealed that new 3D V-Cache will be added to only one chiplet in these CPUs. That means the 64MB cache will be added to one chiplet, while the other chiplet will use the default 32MB cache on the Ryzen 9 7950X3D and 7900X3D. Twitter based user @Andreas Schilling has given a good explanation for it.

The problem is, this means that these processors may not perform as good as initially thought or may have some performance problems. Another question is, how will games use these cores. Thankfully, games might not use more than 8 cores. So will the games come to know which chiplet to use and will games be optimized to handle these new caches.

Tom’s Hardware has a detailed answer on it:

AMD is working with Microsoft on Windows optimizations that will work in tandem with a new AMD chipset driver to identify games that prefer the increased L3 cache capacity and pin them into the CCD with the stacked cache. Other games that prefer higher frequencies more than increased L3 cache will be pinned into the bare CCD. AMD says that the bare chiplet can access the stacked L3 cache in the adjacent chiplet, but this isn’t optimal and will be rare. Yes, the chip with the extra L3 cache will run games at a slower speed, but most games don’t operate at peak clock rates, so you should still get a huge performance benefit.

Famous Twitter based leaker @OneRaichu has explained it in even simple words:

It should be mentioned that some chiplet core related problems have been reported on Ryzen 9 7950X too. The Twitter based handle of benchmarking software CapFrameX has showed how forcing a game to use a single chiplet was giving better FPS in games. In a statement given to Tom’s Hardware, AMD denied any such thing. It is possible that a software update might have fixed it later.

So, if wonder why AMD chose to put all the cache into a single chiplet. It’s for performance reasons and also cost-cutting ones. Putting it into a single chiplet may give better performance and it also means that AMD will not need to add the new and expensive 3D V-Cache tech on two chiplets.

Overview And Impressions

Whatever it is. These new processors are looking impressive. They are much welcomed. If AMD is able to resolve the chiplet and cache usage issue, then these processors could turn out to be great.

The TDP on all these processors have been limited to 120W. So it’s interesting how they will perform.

However, pricing is a big question. If AMD prices them properly, then there’s hardly anything stopping them.

AMD Reveals New Ryzen 7950X3D, 7900X3D, 7800X3D CPUs

The Qi2 wireless charging standard will mandate magnet strength for less slip ‘n slide

While Google Translate, Microsoft Translate, DeepL and many others are ready to translate sentences, paragraphs or even entire texts, they may not be the best options in all cases. ChatGPT's strength is that it focuses on dialog and conversations.

Getting text translated by ChatGPT is straightforward. Just tell it to "translate the text that follows", and it will do so. You should specify the desired target language as well, but that's it.

Here is an example of how that could look like.

"Translate the following text into English: Windows 11 ist der Nachfolger von Windows 10."

ChatGPT translates the text and displays it as its response on the screen.

"Windows 11 is the successor to Windows 10."

ChatGPT may do more than just translate

While it is useful for individual translate jobs, ChatGPT's functionality goes beyond that. Just tell it to translate any text that follows while you communicate with it to a target language.

Anything you type is then translated automatically by ChatGPT; this opens up new possibilities, for instance, of communicating with other (humans) that you don't share a language with.

How good is the translation? As is the case with most machine-based translations: it depends. All tests that I performed produced results that were understandable. Some translations used literal translations for words or phrases, which were not used in the language in that way, but the meaning could be discerned.

You may also instruct ChatGPT to provide you with background information by adding instructions to the translation request. Add "and explain it to me" or "provide background information" to get explanations. You may also instruct it to use a certain tone for the translation, or, and that is certainly a favorite, translate the text as someone else would.

Use "translate the following text into English as a 10-year old child would", to get a different kind of result. If you ever wanted to communicate with your teenage daughter or son using "their language", this may be a way to do it.

Similarly, you could ask it to translate text and then create a poem based on it, or the beginning of a novel.

Interestingly, you may also use it to translate content that it knows already, e.g., Caesar's De Bello Gallico. Again, the option to get more information on certain topics is always just a keystroke away.

Source

A group from Google Research introduced Muse as a tool that’s on par with most current models. However, the researchers went on to say that Muse is significantly more efficient than existing diffusion models like Stable Diffusion, Dalle-E 2 and even Google Parti. However, what is this efficiency scale based on?

During research and extensive testing by GoogleAI, the researchers found that Muse delivers similar quality images, much faster. When evaluated, Muse was compared to Parti-3B and Imagen. Against these competitors, Muse was able to produce images matching them in quality, variety and text alignment. However, Muse stood out as significantly faster than these competitors. 

Muse has a 1.3 second generation time to create images compared to Stable Diffusion which has a 3.7 second generation time. This is a significant difference, making Muse much faster.

The research team was able to achieve this speed in Muse using a compressed discrete latent space and parallel decoding. Regarding text comprehension, Muse uses a frozen T5 language model. This means it fully processes a text prompt instead of focussing on certain words or phrases. This makes it easier to get ahead with using AI-driven tools.

Muse also sports a new architecture that changes how images are adjusted or edited. This new range of image editing applications makes it easier to edit images using text prompts. This makes it possible to make changes in your generated images without needing to use complex masks, but instead you just use prompts. 

During independent testing by other human applicants, Muse images were rated as better suited to text input than Stable Diffusion 1.4. This was the general consensus in 70.6% of testers. Testers also found that Muse is above average when it comes to incorporating predefined words into images. Muse has also shown to be more accurate in its composition than many competitors. This means it’s able to display image elements from the prompt more exactly, ie. three wine bottles or five yellow boxes. 

The Muse team has pointed out that depending on the unique use case there is the ‘potential for harm’. This is not strange when it comes to scientific work on AI systems, especially when it relates to language and images. The tool could be used to reproduce social biases or spread misinformation if used maliciously. As such, the team has decided not to publish the code for Muse. They’ve also held off on releasing a publicly available demo as Muse is a closed model at the moment. If you’re looking for a good image-to-text AI generator in the meantime, why not have a look at ChatGPT? This AI-driven tool can help with more than just images.

Source

We start off with the most interesting demo that was shown off today in the form of a Blender performance demonstration using the BMW scene. In this, AMD showcased its new Ryzen 9 7940HS from the new Ryzen 7040 series lineup beating the Intel Core i7-1280P and the Apple M1 Pro by 45% and 30% respectively. You can see in the screenshot below how the new AMD chip is waiting for Apple and Intel to finish up.

Although it is impressive, bear in mind that these are last-gen products from the competition that AMD compared its new 7940HS with. Here is a performance slide showing Ryzen 7040 series will stack up against Intel and Apple.

Aside from Ryzen 7040, AMD also presented the Ryzen 7045HX series. The HX goes above and beyond 55W and brings in twice the cores over Ryzen 7040. This means Ryzen 7045HX goes all the way up to 16 cores and 32 threads, while Ryzen 7040 series tops out at 8 cores and 16 threads. The 7040 series, however, does have an advantage over 7045HX series in terms of graphics. While the former packs the new RDNA 3-based graphics, the latter comes with RDNA 2. It also features new dedicated Ryzen AI technology, which is the first of that sort in x86 CPUs.

AMD claims its latest and greatest Ryzen 7045HX SKU is up to 169% faster than the best Intel 12th Gen laptop chip.

On the mobile front, AMD had another announcement to make in the form of discrete GPUs. The company released its new Navi 33-based 32CU RX 7600M XT built on RDNA 3.

AMD says that the 7600M XT is up to 31% faster than the Nvidia RTX 3060 for 1080p gaming and 26% faster than its own last-gen stuff on average.

AMD also announced Radeon RX 7700S and Radeon RX 7600S discrete graphics which are meant to be integrated into thinner and lighter notebooks.

AMD Advantage laptops, which will combine the new Ryzen and Radeon chips, are expected to be available starting next month.

Moving on to the desktop side, AMD revealed its Ryzen 7000X3D lineup headlined by the 7950X3D. This time AMD has gone beyond the 8 cores on last-gen 3D V-cache part as 12 cores and 16 cores now get 3D V-cache as well.

In total, AMD will have three 7000X3D chips, the 7950X3D, the 7900X3D and the 7800X3D:

AMD also provided some performance numbers for its new X3D parts.

AMD claims gaming leadership again with its new Ryzen 7950X3D chip over Intel's Core i9-13900K. Meanwhile, the 7800X3D was compared against the 5800X3D, and once again, there are some impressive gains here too.

Perhaps the biggest good news today was the promise of more affordable socket AM5 motherboards and 65W Ryzen 7000 chips, which will help budget shoppers move on to AM5.

AMD Ryzen 7040, 7045, 7000X3D, Radeon RX 7600M XT aim to take down Intel, Apple and Nvidia

‘AI is going to reinvent how you do everything on Windows,’ says Microsoft’s Windows boss