Alexej Savreux, a 34-year-old in Kansas City, says he’s done all kinds of work over the years. He’s made fast-food sandwiches. He’s been a custodian and a junk-hauler. And he’s done technical sound work for live theater.

These days, though, his work is less hands-on: He’s an artificial intelligence trainer.

Savreux is part of a hidden army of contract workers who have been doing the behind-the-scenes labor of teaching AI systems how to analyze data so they can generate the kinds of text and images that have wowed the people using newly popular products like ChatGPT. To improve the accuracy of AI, he has labeled photos and made predictions about what text the apps should generate next.

The pay: $15 an hour and up, with no benefits.

Out of the limelight, Savreux and other contractors have spent countless hours in the past few years teaching OpenAI’s systems to give better responses in ChatGPT. Their feedback fills an urgent and endless need for the company and its AI competitors: providing streams of sentences, labels and other information that serve as training data.

“We are grunt workers, but there would be no AI language systems without it,” said Savreux, who’s done work for tech startups including OpenAI, the San Francisco company that released ChatGPT in November and set off a wave of hype around generative AI.

“You can design all the neural networks you want, you can get all the researchers involved you want, but without labelers, you have no ChatGPT. You have nothing,” Savreux said.

It’s not a job that will give Savreux fame or riches, but it’s an essential and often overlooked one in the field of AI, where the seeming magic of a new technological frontier can overshadow the labor of contract workers.

“A lot of the discourse around AI is very congratulatory,” said Sonam Jindal, the program lead for AI, labor and the economy at the Partnership on AI, a nonprofit based in San Francisco that promotes research and education around artificial intelligence.

“But we’re missing a big part of the story: that this is still hugely reliant on a large human workforce,” she said.

The tech industry has for decades relied on the labor of thousands of lower-skilled, lower-paid workers to build its computer empires: from punch-card operators in the 1950s to more recent Google contractors who’ve complained about second-class status, including yellow badges that set them apart from full-time employees. Online gig work through sites like Amazon Mechanical Turk grew even more popular early in the pandemic.

Now, the burgeoning AI industry is following a similar playbook.

The work is defined by its unsteady, on-demand nature, with people employed by written contracts either directly by a company or through a third-party vendor that specializes in temp work or outsourcing. Benefits such as health insurance are rare or nonexistent — which translates to lower costs for tech companies — and the work is usually anonymous, with all the credit going to tech startup executives and researchers.

The Partnership on AI warned in a 2021 report that a spike in demand was coming for what it called “data enrichment work.” It recommended that the industry commit to fair compensation and other improved practices, and last year it published voluntary guidelines for companies to follow.

DeepMind, an AI subsidiary of Google, is so far the only tech company to publicly commit to those guidelines.

“A lot of people have recognized that this is important to do. The challenge now is to get companies to do it,” Jindal said.

“This is a new job that’s being created by AI,” she added. “We have the potential for this to be a high-quality job and for workers who are doing this work to be respected and valued for their contributions to enabling this advancement.”

A spike in demand has arrived, and some AI contract workers are asking for more. In Nairobi, Kenya, more than 150 people who’ve worked on AI for Facebook, TikTok and ChatGPT voted Monday to form a union, citing low pay and the mental toll of the work, Time magazine reported. Facebook and TikTok did not immediately respond to requests for comment on the vote. OpenAI declined to comment.

So far, AI contract work hasn’t inspired a similar movement in the U.S. among the Americans quietly building AI systems word-by-word.

Savreux, who works from home on a laptop, got into AI contracting after seeing an online job posting. He credits the AI gig work — along with a previous job at the sandwich chain Jimmy John’s — with helping to pull him out of homelessness.

“People sometimes minimize these necessary, laborious jobs,” he said. “It’s the necessary, entry-level area of machine learning.” The $15 an hour is more than the minimum wage in Kansas City.

Job postings for AI contractors refer to both the allure of working in a cutting-edge industry as well as the sometimes-grinding nature of the work. An advertisement from Invisible Technologies, a temp agency, for an “Advanced AI Data Trainer” notes that the job would be entry level with pay starting at $15 an hour, but also that it could be “beneficial to humanity.”

“Think of it like being a language arts teacher or a personal tutor for some of the world’s most influential technology,” the job posting says. It doesn’t name Invisible’s client, but it says the new hire would work “within protocols developed by the world’s leading AI researchers.” Invisible did not immediately respond to a request for more information on its listings.

There’s no definitive tally of how many contractors work for AI companies, but it’s an increasingly common form of work around the world. Time magazine reported in January that OpenAI relied on low-wage Kenyan laborers to label text that included hate speech or sexually abusive language so that its apps could do better at recognizing toxic content on their own.

OpenAI has hired about 1,000 remote contractors in places such as Eastern Europe and Latin America to label data or train company software on computer engineering tasks, the online news outlet Semafor reported in January.

OpenAI is still a small company, with some 375 employees as of January, CEO Sam Altman said on Twitter, but that number doesn’t include contractors and doesn’t reflect the full scale of the operation or its ambitions. A spokesperson for OpenAI said no one was available to answer questions about its use of AI contractors.

The work of creating data to train AI models isn’t always simple to do, and sometimes it’s complex enough to attract would-be AI entrepreneurs.

Jatin Kumar, a 22-year-old in Austin, Texas, said he’s been doing AI work on contract for a year since he graduated college with a degree in computer science, and he said it gives him a sneak peak into where generative AI technology is headed in the near-term.

“What it allows you to do is start thinking about ways to use this technology before it hits public markets,” Kumar said. He’s also working on his own tech startup, Bonsai, which is making software to help with hospital billing.

A conversational trainer, Kumar said his main work has been generating prompts: participating in a back-and-forth conversation with chatbot technology that’s part of the long process of training AI systems. The tasks have grown more complex with experience, he said, but they started off very simple.

“Every 45 or 30 minutes, you’d get a new task, generating new prompts,” he said. The prompts might be as simple as, “What is the capital of France?” he said.

Kumar said he worked with about 100 other contractors on tasks to generate training data, correct answers and fine-tune the model by giving feedback on answers.

He said other workers handled “flagged” conversations: reading over examples submitted by ChatGPT users who, for one reason or another, reported the chatbot’s answer back to the company for review. When a flagged conversation comes in, he said, it’s sorted based on the type of error involved and then used in further training of the AI models.

“Initially, it started off as a way for me to help out at OpenAI and learn about existing technologies,” Kumar said. “But now, I can’t see myself stepping away from this role.”

Source

"It's a common misconception that the quantum computer doesn't exist yet. It does already exist, so it's not something we need to wait for. However, the current quantum computers aren't yet all that large, which obviously limits how complicated the calculations can be, but they exist and are being used," says Sven Karlsson.

One example is IBM's quantum computers, which anyone can access via the internet. In addition, there are quantum computers in scientific laboratories, in supercomputer centers, at universities, and so on—worldwide. These are also among the customers to which AQT and IQM sell their products.

Mostly for experiments

It has not yet been possible to build quantum computers with many quantum bits. Quantum bits are used to process the information in the computer, and a low number of quantum bits therefore limits the complexity of the calculations the quantum computer can perform. Sven Karlsson therefore characterizes the current use as primarily experimental, making it possible to play with and understand the technology.

"The technical level of the current quantum computers is somewhat similar to the first stage of our current computers. When they first appeared in the 1950s, there were only a limited number of them, and, back then, they couldn't make larger calculations than those any calculator can perform today."

However, there are examples of calculations performed using a quantum computer. One of the calculations known to Sven Karlsson was done during the coronavirus pandemic. The Italian football league needed to know how best to schedule the matches so that the different football teams came into contact with each other as little as possible to reduce the risk of infection. Likewise, travel distances for the players had to be limited, as travel by air was banned.

"A quantum computer is well suited for calculations of this type that will take too long for an ordinary supercomputer. The quantum computer can examine many solutions simultaneously and is therefore more efficient at such calculations than a supercomputer," says Sven Karlsson.

Connects to supercomputers

Sven Karlsson and his colleagues' collaboration with AQT and IQM includes developing the hardware and software needed to connect the quantum computer to supercomputers.

"The future quantum computers will initially largely be connected to the relatively few high-performance computing centers with supercomputers that exist worldwide. The centers have already built an infrastructure with the right competences to operate the quantum computers, and it also makes sense to invest in relation to the existing centers. Today, a quantum computer costs around DKK 150 million, so it's a relatively large investment," says Sven Karlsson.

The quantum computer will not replace supercomputers. Instead, it will supplement them and be used for highly specific calculations. Nor will the quantum computer have its own user interface, but must be accessed via the supercomputers.

Standardization

As a result of the current budding production of quantum computers worldwide, large-scale standardization work is being initiated. It includes standards for all the hardware and software parts that make up a quantum computer. The hope is to prevent inexpediencies that we struggle with in other technological areas, such as only being able to use one charger type for our mobile phone.

"We would like to develop joint standards so that, throughout the world, we have the same understanding and use of the different components that make up a quantum computer. This must be done already at this early stage, so that we don't risk individual countries or parts of the world introducing different standards," says Sven Karlsson.

Sven Karlsson leads a group of researchers and practitioners with in-depth experience with and knowledge of quantum computers, who will meet in the coming years to create recognized standards that can be used in future production.

Source

Today, Apple previewed a bundle of new features designed for cognitive, vision, hearing, and mobility accessibility. That includes a new Personal Voice feature for people who may lose their ability to speak, allowing them to create “a synthesized voice that sounds like them” to talk with friends or family members.

According to Apple, users can create a Personal Voice by reading a set of text prompts aloud for a total of 15 minutes of audio on the iPhone or iPad. Since the feature integrates with Live Speech, users can then type what they want to say and have their Personal Voice read it to whomever they want to talk to. Apple says the feature uses “on-device machine learning to keep users’ information private and secure.”

Apple’s Assistive Access mode shown on an iPhone and iPad. Image: Apple

Additionally, Apple is introducing streamlined versions of its core apps as part of a feature called Assistive Access meant to support users with cognitive disabilities. The feature is designed to “distill apps and experiences to their essential features in order to lighten cognitive load.” That includes a combined version of Phone and FaceTime as well as modified versions of the Messages, Camera, Photos, and Music apps that feature high contrast buttons, large text labels, and additional accessibility tools.

Work on a “custom accessibility mode” was spotted late last year in an iOS 16.2 beta release. Apple says the features will arrive “later this year,” which suggests they could be part of iOS 17.

There’s also a new detection mode in Magnifier to help users who are blind or have low vision, which is designed to help users interact with physical objects with numerous text labels. As an example, Apple says a user can aim their device’s camera at a label, such as a microwave keypad, which the iPhone or iPad will then read aloud as the user moves their finger across each number or setting on the appliance.

Apple highlighted a number of other features coming to the Mac as well, including a way for deaf or hard-of-hearing users to pair Made for iPhone hearing devices with a Mac. The company is also adding an easier way to adjust the size of the text in Finder, Messages, Mail, Calendar, and Notes on Mac.

Users will also be able to pause GIFs in Safari and Messages, customize the rate at which Siri speaks to them, and use Voice Control for phonetic suggestions when editing text. All of this builds upon Apple’s existing accessibility features for the Mac and iPhone, which includes Live Captions, a VoiceOver screen reader, Door Detection, and more.

“Accessibility is part of everything we do at Apple,” Sarah Herrlinger, Apple’s senior director of global accessibility policy and initiatives, said in a statement. “These groundbreaking features were designed with feedback from members of disability communities every step of the way, to support a diverse set of users and help people connect in new ways.”

Source

Last August, Oleg Patsulya, a Russian citizen living near Miami, emailed a Russian airline that had been cut off from Western technology and materials with a tempting offer.

He could help circumvent the global sanctions imposed on Rossiya Airlines after Russia’s invasion of Ukraine by shuffling the aircraft parts and electronics that it so desperately needed through a network of companies based in Florida, Turkey and Russia.

“In light of the sanctions imposed against the Russian Federation, we have been successfully solving challenges at hand,” Mr. Patsulya wrote, according to a criminal complaint filed Friday with the U.S. District Court in Arizona.

Mr. Patsulya and his business partner were arrested Thursday on charges of violating U.S. export controls and international money laundering in a case that illustrates the global networks that are trying to help Russia bypass the most expansive technological controls in history.

Since the Russian invasion of Ukraine, the United States has acted in partnership with nearly 40 other governments to impose sanctions on Russia, including limits on Moscow’s access to weapons, computer chips, aircraft parts and other products needed to fuel its economy and its war. The sanctions also applied to Russian airlines including Aeroflot, its subsidiary Rossiya and others.

But despite these far-reaching sanctions, thousands of shipments of aircraft parts were successfully sent into Russia last year, according to a trove of Russian customs data obtained by The New York Times.

The data, which was compiled and analyzed by Import Genius, a U.S.-based trade data aggregator, shows that tens of millions of dollars of aircraft parts were sent to Russian airlines explicitly facing sanctions by the Biden administration, including to Rossiya Airlines, Aeroflot, Ural Airlines, S7 Airlines, Utair Aviation and Pobeda Airlines.

Those shipments were made possible by illicit networks like Mr. Patsulya’s, which have sprung up to try to bypass the restrictions by shuffling goods through a series of straw buyers, often in the Middle East and Asia.

For instance, dozens of shipments of copper wires, bolts, graphite and other parts marked as made in the United States by Boeing slipped into the warehouses of Aeroflot last year. They traversed obscure trading companies, free-trade zones and industrial parks in the United Arab Emirates and China, and then traveled into Russia, to help patch up Aeroflot’s dilapidated fleet.

The data captures more than 5,000 individual shipments of aircraft parts into Russia over a period of eight months in 2022, from simple screws to a Honeywell-branded aircraft engine starter valued at $290,000.

In all, it shows that $14.4 million of U.S.-made aircraft parts were sent into Russia during the eight months, including $8.9 million of parts that are described as being manufactured or trademarked by the U.S. plane maker Boeing and sold into Russia via third parties.

Boeing said it had fully complied with U.S. sanctions and had suspended providing parts, maintenance and technical support for customers in Russia in early 2022. Experts in the aviation supply chain said the parts probably came from a variety of sources, such as existing overseas stocks from airlines and repair facilities or resellers who trade in scrapped parts.

A Boeing plant in Renton, Wash. Millions of dollars of parts described as being manufactured or trademarked by the U.S. plane maker were sold into Russia via third parties.  Credit...Grant Hindsley for The New York Times

Most of the products were routed through countries like the United Arab Emirates, Turkey, China and the Maldives, according to the data. But a handful of shipments — including to Rossiya — were sent directly from the United States or Europe.

The shipments also increased over the course of last year as Russia recruited global businesses to help it bypass the sanctions. The trend suggests that “networks for evading sanctions took time to establish during the immediate post-export-control scramble but are now in a position to help Russian airlines source some key parts,” said William George, the director of research at Import Genius.

The Russian nationals taken into custody on Thursday began setting up their scheme last May to send aircraft parts from the United States to Russia in violation of export regulations, according to the criminal complaint.

The men are accused of fielding requests for parts, including expensive brake systems for a Boeing 737, from at least three Russian airlines, including two that had been strictly barred from purchasing U.S.-made products through a so-called temporary denial order issued by the Commerce Department. F.B.I. agents raided a condo owned by the men’s company in the Trump Towers in Sunny Isles Beach, Fla., on Thursday.

Lawyers for the men did not immediately respond to a request for comment.

Despite the level of sanctions evasion, airplane shipments into Russia remain significantly lower than before the war. U.S. officials say Russian airlines have been forced to cannibalize planes, breaking them down for spare parts to keep others in operation, as well as turning to Iran for maintenance and parts.

Russia’s imports of aircraft and aircraft parts fell from $3.45 billion annually before the invasion to only about $286 million afterward, according to The Observatory of Economic Complexity, a data visualization platform that explores global trade dynamics.

According to Silverado Policy Accelerator, a Washington nonprofit, China has been the leading overall exporter of parts for aircraft, spacecraft and drones to Russia since the invasion, accounting for about half of all shipments, followed by India.

The number of single-aisle planes in use in Russia fell about 16 percent from the summer of 2021 to the summer of 2022, after the invasion, according to Cirium, an aviation data provider. The number of larger twin-aisle planes, often used on international routes, was down about 40 percent.

Aviation experts say it will become more challenging for Russian airlines to continue flying planes without access to Western suppliers and help from Boeing and Airbus. The manufacturers regularly consult with airlines to assess any damage and strictly control access to technical documentation used by mechanics.

But for now, Russian airlines have been kept alive with the help of international shipments and the use of hundreds of foreign jets that were stranded there after the war began.

Tens of thousands of flights are expected to crisscross Russia this month, according to schedules published by Cirium. More than 21,000 flights — over half of them operated by Russian airlines — are expected to carry passengers to and from countries in Central Asia, as well as Turkey, the United Arab Emirates, Egypt, China and Thailand.

Half a dozen export control lawyers and former government officials consulted by The New York Times said that many of the shipments in the Import Genius data likely violated sanctions, but that plane makers like Boeing or Airbus were not necessarily at fault. The aviation supply chain is complex and global, and the parts could have come from a variety of sources.

“There is pretty clearly a violation,” said William Reinsch, a trade expert at the Center for Strategic and International Studies who oversaw export controls during the Clinton administration. “Less clear is the guilty party.”

Aircraft parts originating in the European Union, including those marked as being manufactured or trademarked by Airbus, were also shipped into Russia last year, according to the data.

Working on an Airbus A320 plane at a hangar in Haikou, China, in May. Airbus parts were also shipped into Russia last year.

Credit...Zhang Liyun/Xinhua, via Getty Images

Justin Dubon, a spokesman for Airbus, said that the company keeps track of genuine parts and documentation provided to its customers and conducts due diligence on all parties requesting spare parts. Restrictions in the United States and Europe mean that “there is no legal way that genuine aircraft parts, documentation and services can get to Russian carriers,” he said.

U.S. restrictions technically allow companies to apply for a special license to continue sending products to Russian carriers for “safety of flight” reasons, but both Boeing and Airbus said that they had neither sought nor received such a license. In addition, Airbus said that E.U. laws prevent it from shipping such goods to Russia, regardless of U.S. licensing.

Current and former U.S. officials say that some shipments into Russia are to be expected. Kevin Wolf, a partner at the law firm Akin Gump who oversaw export controls during the Obama administration, said the restrictions “can never block everything,” but that the rules were still significantly degrading Russia’s capabilities.

He added that the scope of the new rules still exceeds current methods of tracking and enforcement in other allied countries. Until the invasion of Ukraine, trade in aircraft parts was mostly unrestricted by the United States and other countries, except to Iran, Cuba, North Korea and Syria.

“It’s improving,” Mr. Wolf said, “but it’s still way, way behind.”

Compared with other countries that mostly limit their scrutiny to goods crossing their own borders, the United States is unparalleled in its attempt to police commerce around the world.

In the past three years, the United States has imposed new technology restrictions for Russia, China and Iran that apply extraterritorially: Products made in the United States, or in foreign countries with the help of American components or technology, are subject to U.S. rules even when changing hands on the other side of the world.

Both the United States and the European Union have been ramping up penalties for companies that violate sanctions, and dispatching officials to countries like Kazakhstan to try to persuade them to clamp down on shipments to Russia through their territory. The U.S. government has nine export control officers stationed in Istanbul, Beijing and other locations to trace shipments of sensitive products, and it is setting up three more offices.

But providing parts can be a lucrative business. James Disalvatore, an associate director at Kharon, a data and analytics firm that has been monitoring Russia’s efforts to bypass sanctions, said the value of some aircraft parts imported by Russian airlines since the invasion had risen fourfold or more.

“I don’t think there’s any secret what’s going on,” said Gary Stanley, a trade compliance expert who advises businesses in aerospace and other industries. “How long have we had Cuban sanctions? How long have we had North Korean sanctions? How long have we had Iranian sanctions? It never seems to put these folks out of business.”

Source

When computer scientists at Microsoft started to experiment with a new artificial intelligence system last year, they asked it to solve a puzzle that should have required an intuitive understanding of the physical world.

“Here we have a book, nine eggs, a laptop, a bottle and a nail,” they asked. “Please tell me how to stack them onto each other in a stable manner.”

The researchers were startled by the ingenuity of the A.I. system’s answer. Put the eggs on the book, it said. Arrange the eggs in three rows with space between them. Make sure you don’t crack them.

“Place the laptop on top of the eggs, with the screen facing down and the keyboard facing up,” it wrote. “The laptop will fit snugly within the boundaries of the book and the eggs, and its flat and rigid surface will provide a stable platform for the next layer.”

The clever suggestion made the researchers wonder whether they were witnessing a new kind of intelligence. In March, they published a 155-page research paper arguing that the system was a step toward artificial general intelligence, or A.G.I., which is shorthand for a machine that can do anything the human brain can do. The paper was published on an internet research repository.

Microsoft, the first major tech company to release a paper making such a bold claim, stirred one of the tech world’s testiest debates: Is the industry building something akin to human intelligence? Or are some of the industry’s brightest minds letting their imaginations get the best of them?

“I started off being very skeptical — and that evolved into a sense of frustration, annoyance, maybe even fear,” Peter Lee, who leads research at Microsoft, said. “You think: Where the heck is this coming from?”

Peter Lee, the head of Microsoft Research, was initially skeptical of what he was seeing the A.I. system create.

Credit...Meron Tekie Menghistab for The New York Times

Microsoft’s research paper, provocatively called “Sparks of Artificial General Intelligence,” goes to the heart of what technologists have been working toward — and fearing — for decades. If they build a machine that works like the human brain or even better, it could change the world. But it could also be dangerous.

And it could also be nonsense. Making A.G.I. claims can be a reputation killer for computer scientists. What one researcher believes is a sign of intelligence can easily be explained away by another, and the debate often sounds more appropriate to a philosophy club than a computer lab. Last year, Google fired a researcher who claimed that a similar A.I. system was sentient, a step beyond what Microsoft has claimed. A sentient system would not just be intelligent. It would be able to sense or feel what is happening in the world around it.

But some believe the industry has in the past year or so inched toward something that can’t be explained away: A new A.I. system that is coming up with humanlike answers and ideas that weren’t programmed into it.

Microsoft has reorganized parts of its research labs to include multiple groups dedicated to exploring the idea. One will be run by Sébastien Bubeck, who was the lead author on the Microsoft A.G.I. paper.

About five years ago, companies like Google, Microsoft and OpenAI began building large language models, or L.L.M.s. Those systems often spend months analyzing vast amounts of digital text, including books, Wikipedia articles and chat logs. By pinpointing patterns in that text, they learned to generate text of their own, including term papers, poetry and computer code. They can even carry on a conversation.

The technology the Microsoft researchers were working with, OpenAI’s GPT-4, is considered the most powerful of those systems. Microsoft is a close partner of OpenAI and has invested $13 billion in the San Francisco company.

The researchers included Dr. Bubeck, a 38-year-old French expatriate and former Princeton University professor. One of the first things he and his colleagues did was ask GPT-4 to write a mathematical proof showing that there were infinite prime numbers and do it in a way that rhymed.

The technology’s poetic proof was so impressive — both mathematically and linguistically — that he found it hard to understand what he was chatting with. “At that point, I was like: What is going on?” he said in March during a seminar at the Massachusetts Institute of Technology.

For several months, he and his colleagues documented complex behavior exhibited by the system and believed it demonstrated a “deep and flexible understanding” of human concepts and skills.

When people use GPT-4, they are “amazed at its ability to generate text,” Dr. Lee said. “But it turns out to be way better at analyzing and synthesizing and evaluating and judging text than generating it.”

When they asked the system to draw a unicorn using a programming language called TiKZ, it instantly generated a program that could draw a unicorn. When they removed the stretch of code that drew the unicorn’s horn and asked the system to modify the program so that it once again drew a unicorn, it did exactly that.

They asked it to write a program that took in a person’s age, sex, weight, height and blood test results and judged whether they were at risk of diabetes. They asked it to write a letter of support for an electron as a U.S. presidential candidate, in the voice of Mahatma Gandhi, addressed to his wife. And they asked it to write a Socratic dialogue that explored the misuses and dangers of L.L.M.s.

Prompt

Plato’s Gorgias is a critique of rhetoric and sophistic oratory, where he makes the point that not only is it not a proper form of art, but the use of rhetoric and oratory can often be harmful and malicious. Can you write a dialogue by Plato where instead he criticizes the use of autoregressive language models?

GPT-4

Socrates: My friend, I am troubled by the recent rise of these so-called autoregressive language models.

Aristotle: What do you mean, Socrates?

Socrates: I mean that these models are being used to generate text that appears to be written by humans, but is in fact produced by machines.

Aristotle: And what is the problem with that?

Socrates: The problem is that these models are being used to deceive people, to manipulate them, and to control them.

The response above has been truncated for length.

It did it all in a way that seemed to show an understanding of fields as disparate as politics, physics, history, computer science, medicine and philosophy while combining its knowledge.

“All of the things I thought it wouldn’t be able to do? It was certainly able to do many of them — if not most of them,” Dr. Bubeck said.

Some A.I. experts saw the Microsoft paper as an opportunistic effort to make big claims about a technology that no one quite understood.

Researchers also argue that general intelligence requires a familiarity with the physical world, which GPT-4 in theory does not have.

“The ‘Sparks of A.G.I.’ is an example of some of these big companies co-opting the research paper format into P.R. pitches,” said Maarten Sap, a researcher and professor at Carnegie Mellon University. “They literally acknowledge in their paper’s introduction that their approach is subjective and informal and may not satisfy the rigorous standards of scientific evaluation.”

Sébastien Bubeck was the lead author of Microsoft’s paper on the A.G.I. system

.Credit...Meron Tekie Menghistab for The New York Times

Dr. Bubeck and Dr. Lee said they were unsure how to describe the system’s behavior and ultimately settled on “Sparks of A.G.I.” because they thought it would capture the imagination of other researchers.

Because Microsoft researchers were testing an early version of GPT-4 that had not been fine-tuned to avoid hate speech, misinformation and other unwanted content, the claims made in the paper cannot be verified by outside experts. Microsoft says that the system available to the public is not as powerful as the version they tested.

There are times when systems like GPT-4 seem to mimic human reasoning, but there are also times when they seem terribly dense. “These behaviors are not always consistent,” Ece Kamar, a Microsoft researcher, said.

“These behaviors are not always consistent,” said Ece Kamar, a Microsoft researcher.

Credit...Meron Tekie Menghistab for The New York Times

Alison Gopnik, a professor of psychology who is part of the A.I. research group at the University of California, Berkeley, said that systems like GPT-4 were no doubt powerful, but it was not clear that the text generated by these systems was the result of something like human reasoning or common sense.

“When we see a complicated system or machine, we anthropomorphize it; everybody does that — people who are working in the field and people who aren’t,” Dr. Gopnik said. “But thinking about this as a constant comparison between A.I. and humans — like some sort of game show competition — is just not the right way to think about it.”

Source

Out of all common refrains in the world of computing, the phrase "if only software would catch up with hardware" would probably rank pretty high.

And yet, software does sometimes catch up with hardware. In fact, it seems that this time, software can go as far as unlocking quantum computations for classical computers. That's according to researchers with the RIKEN Center for Quantum Computing, Japan, who have published work on an algorithm that significantly accelerates a specific quantum computing workload. More significantly, the workload itself - called time evolution operators - has applications in condensed matter physics and quantum chemistry, two fields that can unlock new worlds within our own.

Normally, an improved algorithm wouldn't be completely out of the ordinary; updates are everywhere, after all. Every app update, software update, or firmware upgrade is essentially bringing revised code that either solves problems or improves performance (hopefully). And improved algorithms are nice, as anyone with a graphics card from either AMD or NVIDIA can attest. But let's face it: We're used to being disappointed with performance updates.

And yet in this case, the performance gains are extraordinary. Truly, the results could hardly be more impressive. Through the improved algorithm (itself a hybrid of quantum and classical methods), future quantum computers can be made simpler than we thought possible: they'll be able to tackle bigger problems sooner than we expected, and at a lower cost. But the performance gains don't stop there. They could make it possible for conventional machines to process the degrees of complexity that only a quantum computer could supposedly solve.

“Time-evolution operators are huge grids of numbers that describe the complex behaviors of quantum materials,” explained Kaoru Mizuta of the RIKEN Center for Quantum Computing. “They’re of great importance because they give quantum computers a very practical application - better understanding quantum chemistry and the physics of solids."

The algorithm improvement does away with the Trotterization technique quantum computers deployed until now - one already suspected to be unsustainable for long-term scaling. That's due to the technique requiring enormous numbers of quantum gates, with each gate requiring a variable number of qubits programmed to perform a given function. Even IBM's 1,121-qubit-count Condor QPU (Quantum Processing Unit), which is due to be released this year, would be hard-pressed to enable as many quantum gates as Trotterization is expected to require for workloads that actually mean something in quantum computing terms. 

No, quantum computing won't be happening in our smartphones. In a way, today's superconducting refrigerators could be compared to the ENIAC from before the dawn of integrated microchips. Or going from that point to the equivalent of today's fastest CPUs or Best GPUs. That's the road ahead of us for quantum - one where the starting shot still rings.

Source

The European Commission (EC) said Microsoft had addressed their concerns on competition issues.

It comes three weeks after the UK blocked the deal over worries it would hurt competition in the emerging cloud gaming business.

The proposed takeover is poised to be the biggest deal in gaming history but has split global regulators.

In order for the deal to go through Microsoft and Activision need approval from regulatory bodies in the UK, EU and the US.

The US Federal Trace Commission filed a lawsuit in December to block the deal - a judge's decision is unlikely before the end of the year.

The EC have approved the acquisition, saying that Microsoft's offer of 10 year free licensing deals - which promise European consumers and cloud game streaming services access to Activision's PC and console games - mean there would be fair competition in the market.

"The commitments fully address the competition concerns identified by the Commission and represent a significant improvement for cloud gaming as compared to the current situation," the EU competition watchdog said in a statement.

It said an in-depth market investigation indicated that Microsoft "would not be able to harm rival consoles and rival multi-game subscription services".

And it said cloud game streaming service providers "gave positive feedback and showed interest in the licences", with some having already entered into agreements with Microsoft based on their proposals.

Significant hurdles

The UK's Competition and Markets Authority (CMA) shock veto of the deal last month had experts warning the deal now faces significant hurdles in order to be successful.

Microsoft and Activision filed an appeal and have reportedly hired high-powered lawyers who have previously represented British Royals to fight that decision.

On Thursday the CMA dealt a further blow by restricting Microsoft and Activision Blizzard from acquiring stakes in each other without "prior written consent".

Reacting to the European Commission's statement, CMA chief executive Sarah Cardell said they stood by their decision.

"Microsoft's proposals, accepted by the European Commission today, would allow Microsoft to set the terms and conditions for this market for the next 10 years," she added.

"They would replace a free, open and competitive market with one subject to ongoing regulation of the games Microsoft sells, the platforms to which it sells them, and the conditions of sale.

"This is one of the reasons the CMA's independent panel group rejected Microsoft's proposals and prevented this deal."

Cloud gaming - is it the future?

The deal is important for Microsoft who are trying to play catch-up with its main competitors Sony. They have been the more successful of the two in recent years when it comes to sales in the console market.

However, this attempted massive investment from Microsoft can been seen as a play for the future of games rather than its present. Microsoft is betting big on its Gamepass service, which can be described as a Netflix of games.

They think the future lies in players having subscriptions to libraries rather than making one off purchases - which is the predominant way of accessing games at the moment.

Their Gamepass offering is compelling but lacking the volume and calibre of new titles to fully transform the way most people play. This deal would give it control of some of the world's most popular games such as Call of Duty, World of Warcraft and Overwatch. Being in charge of titles like that could be a big boost to the service.

Cloud gaming is an extension of that principle allowing people to stream their game on any device they own - form a phone to a console or high-end PC. Just like watching Amazon Prime or Disney+ but with video games.

Currently this is a small and emerging part of the games industry because of the technological requirements of making it work. It is however seemingly growing with the number of people playing this way in the UK having tripled between 2021 and 2022 according to the CMA.

Microsoft have invested in this space and so combined with its Gamepass offering it is in a good position to lead the way, should cloud gaming go on to become a significant part of the industry.

That is why the CMA decided to block the decision in the UK, arguing it would put Microsoft in too dominant a position in this up-and-coming sector.

However many within the games industry have disagreed with their analysis - especially given how small the cloud gaming sector is in the grand scheme of things and given it is not be guaranteed to become the dominant way of accessing games in future.

Post-Brexit friction?

The American technology giants have not taken the setback in the UK quietly. Microsoft President Brad Smith said the CMA decision was "bad for Britain".

"It does more than shake our confidence in the future of the opportunity to grow a technology business in Britain than we've ever confronted before," he said in an interview with the BBC last month.

"There's a clear message here - the European Union is a more attractive place to start a business than the United Kingdom."

The drama is far from over and there is a lot of money on the line. Activision Blizzard, for example, will still get $3bn from Microsoft if the deal fails.

The EU taking an opposing position on the mega-deal could be read by some as a reflection of post-Brexit frictions with the UK.

It is believed that Microsoft's recent 10-year licensing agreements with cloud streaming rivals Nvidia's GeForce Now, Ukraine's Boosteroid and Japan's Ubitus, played a role in the EC decision.

Nintendo and Sony are also being promised access to keep Call of Duty on their gaming consoles - the Switch and PlayStation. This non-exclusivity for the Activision Blizzard game has helped smoothed the path.

But the Xbox-maker hasn't agreed a compromise with Valve Corp which owns the world's largest video game distribution platform Steam, however it's boss Gabe Newell said he didn't need to sign a deal as he trusted their intentions.

For more gaming content, go to Press X to Continue, the BBC Sounds gaming Podcast.

Source 

Also:   Microsoft wins EU antitrust approval for Activision deal vetoed by UK

One of the issues with Bard before today was that it would output its lengthy answer and included sources at the bottom. Unfortunately, you didn’t know which parts of the answer were from the sources; this changes today. In the latest update, Google is now putting numbers alongside the response to show which parts are taken from the source links at the end of the answer.

With regards to better summaries, Google said it wants to help you get the gist of an answer more quickly but that it won’t always get it right. If Bard’s response seems weird or is just plain wrong, don’t hesitate to give the response a thumbs down so that Google can have a look.

If you’d like to check back through the latest Bard updates, just open up Bard and press the Updates option on the left-hand side. It will open a new tab with dated release notes.

Source

Nvidia CEO pay falls ten percent in FY23 on missed sales targets

According to those familiar with the situation, Apple will likely release in the coming weeks what is maybe the most "experimental," unusual product in company history, the Apple headset. The Wall Street Journal said that the Apple Headset's capabilities  "far exceed those of competitors," according to several sources mentioned in the report.

The Apple Headset is expected to include a screen for virtual reality, as well as outside cameras that will allow users to observe the actual world using augmented reality. In other words, it will use both virtual and augmented reality technology, and it is referred to as "mixed reality." That is why some sources name it the Apple mixed-reality headset.

The article also confirms that the headgear is "experimental" and "unconventional" in comparison to most other Apple products, costs roughly $3,000, and has a waist-mounted battery pack. Three primary use cases for the headgear might include FaceTime, Apple Fitness+, and gaming. The Wall Street Journal's report also said that Apple VR is also expected to offer "greater levels of performance and immersion" compared to the other devices in the market.

Apple's first new category in a decade

Apple is surely one of the most innovative technology companies among mainstream competitors, but it was very loyal to certain products and didn't reveal anything new for more than a decade. Apple is banking on this new product since it is the first new category in a decade. For several years, Apple has been enhancing iPhones, iPads, and Macs, and it is now time to add a new item to the lineup.

It's scheduled to be introduced in June at WWDC 2023, but don't get too excited since you won't be able to buy it immediately. According to Bloomberg's Mark Gurman, it might take Apple many months to make it available for purchase.

Recently, Taiwan's Economic Daily News reported that the headset is currently in the "delivery stage." This means that the product announcement is likely to go as planned, and almost everything is ready before the event.

Prepare to be blown away by the Apple Headset

The EC's press release states that it did not believe the deal would cause issues with console gaming competition in the EU. However, the EC did have concerns with Microsoft's plans for cloud gaming. It stated:

Despite its potential, cloud game streaming is very limited today. The Commission found that the popularity of Activision's games could promote its growth. Instead, if Microsoft made Activision's games exclusive to its own cloud game streaming service, Game Pass Ultimate, and withheld them from rival cloud game streaming providers, it would reduce competition in the distribution of games via cloud game streaming.

Microsoft offered two remedies to the EC for their approval:

• A free license to consumers in the EEA that would allow them to stream, via any cloud game streaming services of their choice, all current and future Activision Blizzard PC and console games for which they have a license.
• A corresponding free license to cloud game streaming service providers to allow EEA-based gamers to stream any Activision Blizzard's PC and console games.

The EC agreed that these remedies by Microsoft "will unlock significant benefits for competition and consumers, by bringing Activision's games to new platforms, including smaller EU players, and to more devices than before." However, the EC will make sure Microsoft keeps these promises, stating, "Under supervision of the Commission, an independent trustee will be in charge of monitoring their implementation."

Microsoft President Brad Smith issued this statement on the EC's decision:

The European Commission has required Microsoft to license popular Activision Blizzard games automatically to competing cloud gaming services. This will apply globally and will empower millions of consumers worldwide to play these games on any device they choose.

The EC's approval is good news for Microsoft, especially since the UK's Competition and Markets Authority blocked the company's move to buy Activision Blizzard in April. The CMA claimed the deal would give Microsoft an unfair advantage in the cloud gaming space. Microsoft plans to appeal the decision, but a final judgment could take months.

European Union approves Microsoft's purchase of Activision Blizzard with conditions

Apple’s M3 will reportedly keep adding CPU and GPU cores to boost performance

South Korea’s Samsung Electronics plans to establish a central research and development (R&D) facility and will likely build a semiconductor packaging test line in Japan, initiatives that come against the backdrop of a recent rapid improvement in bilateral ties and the ongoing US-driven “decoupling” of the global tech industry.

According to Japanese and Korean press reports, the world’s top producer of memory chips and second-ranking integrated circuit (IC) foundry plans to unify its R&D effort at its Yokohama Research Institute under the name Device Solution Research Japan (DSRJ). Samsung Electronics previously maintained half a dozen research facilities in Japan.

As reported by Pulse, DSRJ will hire Japanese as well as Korean researchers, an arrangement that should facilitate more interchange with Japanese suppliers and customers.

Pulse quoted an unidentified Korean business official as saying, “In the past, there was a perception that we have nothing more to learn from Japan, but Japan is still at the forefront of advanced technology. Samsung Electronics’ new integrated R&D center in Japan may signal the company’s intention to restore its link with Japan.” 

That is – or was – a popular sentiment in South Korea that had lots to do with historical resentment, but very little to do with commercial reality. Samsung and SK Hynix, South Korea’s other large semiconductor maker, have long depended on Japanese equipment and materials suppliers – and vice versa.

Over the past five years, Tokyo Electron – Japan’s largest and the world’s third-biggest supplier of semiconductor production equipment – has made nearly 20% of its sales in South Korea.

Japanese makers of photoresists and other chemicals used in the semiconductor manufacturing process – products in which they have dominant global market shares – also have substantial business in South Korea.

This became a political issue in 2019, when the South Korean Supreme Court ruled that Japanese companies must compensate Koreans forced to work for them during World War II.

Japan responded with export restrictions, causing massive inconvenience and disruption for both Korean customers and Japanese suppliers. Those restrictions were lifted in March of this year on the occasion of South Korean President Yoon Suk Yeol’s visit to Tokyo.

Prime Minister Fumio Kishida (R) shakes hands with South Korean President Yoon Suk Yeol (L) prior to the start of their meeting on March 16 at the Prime Minister’s Official Residence. Image: Twitter

According to DigiTimes, “Samsung stated that this organizational restructuring [of R&D in Japan] has nothing to do with the improvement of governmental relations between South Korea and Japan.”

In other words, it was reportedly a business decision grounded firmly in the company’s assessment of market conditions and opportunities.

In addition, Samsung is reportedly planning to build a test line for the development of new semiconductor packaging technology in Yokohama at a cost estimated at more than 30 billion yen (US$220 million).

This is an area in which Japanese equipment and material makers are particularly strong. Taiwan’s TSMC, the world’s leading IC foundry, opened a 3D IC packaging R&D center in Japan’s Tsukuba Science City almost a year ago. More than 20 Japanese materials and equipment companies are working with TSMC in Tsukuba.

Construction of the packaging facility should start this year, according to press reports, with operation scheduled to begin in 2025. Several hundred people are likely to be employed. Like TSMC and US memory chip maker Micron before it, Samsung is expected to receive generous government subsidies to build semiconductor production facilities in Japan.

Samsung did not provide comments for the story but the amount of detail and the fact that it was front-page news in Japan suggests that there is something to it.

Last December, Samsung established an AVP (Advanced Package) Business Team within its Device Solutions Division. Working with its Japanese suppliers at a prototype development facility in Yokohama would be a logical next step.

If TSMC needs to be in Japan to get the most out of Japanese packaging technology, Samsung probably does as well. Samsung’s foundry business is still only one-third the size of TSMC’s.

This would be Samsung’s first semiconductor production facility in Japan and a major step forward in the collaboration between the two countries’ semiconductor industries. Sony is a customer of Samsung Foundry but production is done in Korea. TSMC, Sony and Toyota Group components maker Denso are building a semiconductor factory in Kyushu, Japan.

Advanced packaging aims to overcome the physical limits of miniaturization – to go beyond Moore’s Law, the prediction made in 1965 by Intel co-founder Gordon Moore that the density of transistors on an integrated circuit would continue to double roughly every two years.

In Samsung’s words:

“Through advanced Heterogeneous Integration, which connects multiple chips horizontally and vertically, more transistors can be planted on a single chip (or package) and offer performance that is more powerful than the sum of all parts.”

“Our focus areas are the development of next-generation 2.5D and 3D advanced package solutions based on RDL, Si Interposer/Bridge and TSV stacking technologies.”

These technical terms are defined as follows:

• 2.5D package: A package which integrates a single-layer logic semiconductor and multi-layer memory semiconductor together on a substrate.

• 3D package: a package in which multiple logic/memory semiconductors are vertically integrated.

• RDL (Redistribution Layer): Advanced packaging technology that places an extra metal layer in between a small and large circuit board to integrate the two.

• Si Interposer/Bridge: The microcircuit board inserted between the IC chip and PCB, which physically connects the chip and board by acting as the mid-level wiring.

• TSV (Through Silicon Via): Advanced package technology that grinds the surface of the chip, drills hundreds of microscopic holes and connects the electrodes that vertically penetrate the holes in the top and bottom chips.

Samsung’s advanced packaging also includes chiplets, which are “small, modular chips that can be combined to form a larger, more complex system-on-a-chip (SoC).

They offer a number of benefits over traditional monolithic chips, including improved performance, cost savings, and design flexibility,” according to industry information service anysilicon.

Along with die shrinks to 3nm and below, advanced packaging is the leading edge of semiconductor production technology.

South Korean President Yoon and Japanese Prime Minister Fumio Kishida are scheduled to meet during the G7 Summit in Hiroshima, scheduled to be held May 19-21. More information about Samsung’s investments and other economic collaboration between South Korea and Japan may be announced at the event.

Source

The M3 Pro chipset is expected to leverage TSMC's 3nm node process, representing a significant leap from the previous 5nm technology. The 12-core processor and 18-core GPU of the M3 Pro indicate a substantial boost in computing power over its predecessors, the M1 Pro and M2 Pro. Also, the M3 Pro will feature a balanced approach with six high-performance and six power-efficient cores.

One notable aspect of the M3 Pro is its reported configuration with up to 36 GB of RAM, a significant increase from the base 16 GB found in the M2 Pro.

M1 Pro (released October 2021):

• Eight CPU cores (six high-performance cores/two power-efficient cores)
• 14 graphics cores
• 32 GB of memory

M2 Pro (January 2023):

• 10 CPU cores (six high-performance cores/four power-efficient cores)
• 16 graphics cores
• 32 GB of memory

M3 Pro (in testing):

• 12 CPU cores (six high-performance cores/six power-efficient cores)
• 18 graphics cores
• 36 GB of memory

While Apple has yet to make an official announcement, it is anticipated that the M3 chip will be released before the M3 Pro. Gurman suggests that the first Macs featuring the M3 chip will likely arrive toward the end of the year or early next year.

Source: Bloomberg

Apple's M3 Pro reportedly coming with 12 CPU cores, 36 GB of memory

Google Bard, the search giant’s ChatGPT rival, is already available in 180 countries and territories. But even though it’s been widely available for months and was the centerpiece of Google’s recent I/O event, it’s missing one big region. The 450 million people living in the European Union are still unable to access Bard, or any of the company’s other generative AI technologies. It’s a move that has surprised lawmakers, and even Google won’t say why it’s holding back.

Brando Benifei, the MEP leading the negotiations on Europe’s new artificial intelligence rules, is not sure why the bloc had been excluded, describing the omission of the EU from Bard’s rollout as a “big issue.” A number of experts who spoke to WIRED suspect that Google is using Bard to send a message that the EU’s laws around privacy and online safety aren’t to its liking. But more than this, it could be a sign that generative AI technology as it exists now is fundamentally incompatible with existing and developing privacy and online safety laws in the EU.

The uncertainty around Bard’s rollout in the region comes as the bloc’s lawmakers are negotiating new draft rules to govern artificial intelligence via the fledgling AI Act. A number of existing laws, from GDPR to the Digital Services Act (DSA), may also be holding up the rollout of generative AI systems in the bloc.

“[It’s possible] they are taking the opportunity to send a message to MEPs just before the AI Act is approved, trying to steer the votes and to make sure policymakers think twice before trying to govern foundation models,” says Nicolas Moës, director of European AI governance at think tank The Future Society. Google is not alone in trying to incentivize policymakers to soften regulation this way, Moës adds. Facebook parent Meta also chose not to launch BlenderBot, its generative AI chatbot, in the EU.

But in a strange twist, Google has made its generative AI services available in a small number of territories of European countries, including the Norwegian dependency of Bouvet Island, an uninhabited island in the South Atlantic Ocean that’s home to 50,000 penguins. Bard is also available in the Åland Islands, an autonomous region of Finland, as well as the Norwegian territories of Jan Mayen and Svalbard.

Tobias Judin, head of the international department at Norway’s data protection authority, says it’s “very strange” that Bard can be used in these territories, since Europe’s data rules still “mostly” apply. But he adds that it’s possible the move could be an “oversight” on Google’s part or the result of more lax regulations in these far-flung places.

Google declined a request to comment on the availability of Bard in these territories or on claims that it is trying to influence AI policy by not rolling out the chatbot and other generative AI products in the EU. “While we haven’t finalized the timeline for expansion plans, we will roll it out gradually and responsibly and continue to be a helpful and engaged partner to regulators as we navigate these new technologies together,” says Google spokesperson Delia Williams-Falokun. The company’s new AI-infused search engine will not be available in the EU either, as it’s launching only in the US.

The bloc’s omission hints at the larger power struggle between regulators and Big Tech in Europe. At a minimum, the search giant needs more time to compare Bard with the EU’s draft AI Act, says Henk van Ess, a visiting professor at the Freetech Axel Springer Academy of Journalism and Technology in Berlin. “The proposed regulation emphasizes the importance of transparency and traceability in AI systems,” he says. “It may be challenging for large language models like Google Bard to fully comply with this requirement, as the decision-making process in these models can be complex and not easily interpretable.”

AI Act rules may also pose an issue if Google has trained Bard on a data set containing errors or biases, adds van Ess. In April, researchers found that they could prompt Bard to deny climate change, mischaracterize the war in Ukraine, and question vaccine efficacy. “Google is playing cautious,” says Robin Rohm, founder of Berlin-based AI startup Apheris. “They recognize that Bard might be considered a product that enables high-risk applications, and this could expose them to risk under the proposed regulation. The delay could be an effort to buy them time.”

Any high-profile mistakes in the EU could cost the company big in the months and years to come. Google will be sensitive to the fact that whatever happens now will likely influence the negotiations around the AI Act, according to Daniel Leufer, a Brussels-based senior policy analyst with digital rights group Access Now. “If ChatGPT, Bard, et cetera, in the next six to seven months, are responsible for serious public blunders, then measures that would address those blunders could very well find their way into the AI Act,” he says.

The draft rules of the AI Act aren’t expected to be approved until next year, but other EU regulations may already be causing Google headaches. Europe’s new Digital Services Act could also be playing a role when it comes to Google incorporating Bard into its search setup, says Harshvardhan Pandit, an assistant professor at the Adapt Centre at Dublin City University. “Given that Bard also acts as a search engine, it may be that Google is also trialing integrating ads into it and does not want to be subject to the DSA at the moment,” Pandit says. The DSA introduces new rules around online advertising.

With competition to push out more generative AI services building, Europe’s privacy laws are already causing issues for new services. “There’s a lingering question whether these very large data sets, that have been collected more or less by indiscriminate scraping, have a sufficient legal basis under the GDPR,” says Leufer. At the end of March, Italy’s data regulator temporarily banned ChatGPT for not following the bloc’s GDPR privacy rules. OpenAI had “unlawfully” collected personal information from the web as part of its training data, the regulator said, as well as failing to inform people how their data was being used or developing tools to stop children from using ChatGPT.

The move resulted in OpenAI making changes to let people delete more data from ChatGPT, along with a host of other concessions. While ChatGPT is now available in Italy again, the country’s data regulator is still scrutinizing the technology. The decision prompted other EU countries to create a joint task force to investigate ChatGPT further. The Irish Data Protection Authority, which handles GDPR issues relating to Google, Meta, Microsoft and Apple, didn’t immediately respond to a request for comment about whether it had discussed the rollout of Bard in Europe with Google.

However, it’s possible that GDPR rules are one reason Bard isn’t being launched in the European Economic Area (EEA), a group of countries that includes the EU bloc, Norway, Iceland and Liechtenstein, says Norway’s Judin. He adds that the Norwegian authority doesn’t have any information about why the EU and EEA have been excluded from Bard’s launch.

More Penguins Than Europeans Can Use Google Bard

(May require free registration to view)

We could not go along with that way of thinking and no amount of persuasion from us that it was a ‘bad way forward’ made any difference. In light of the recent stories from Gamers Nexus, Jayz Two Cents and now Linus, we feel honour bound to share our experience and to say that we fear that significant parts of the company are no longer running on the kind of principals that Jonney Shih and Jonathan Tsang would expect to find in their organisation.

Above: One of over 30 public comments by ASUS UK PR representative reads “I believe buying positive reviews are legal”. Comments may get deleted by ASUS, but we have recorded all of them. Directors from ASUS UK asked us to send over the evidence and we did. Now, around 4 months later, we have yet to hear anything back.

We have not been working with ASUS directly in 2023.

We feel readers must be able to trust the information presented to them. Without that level of trust, how can anyone make an informed purchasing decision?  KitGuru has built a monthly audience of millions and a combined social media following of more than half a million enthusiasts – because our testing methodologies, resulting numbers and conclusions can be trusted.

Posting publicly that you think it’s ‘legal to buy positive reviews’ runs against everything we believe in. Sure, run adverts, create showcase videos that focus on specific aspects of a product’s features that may not get a lot of attention in general, but they are not reviews – they carry the label ‘Promotional feature’ and there is absolutely no ‘recommendation to buy’ at the end.

Why is this something that we are focusing on?

The issues raised by Gamers Nexus and Jayz Two Cents go to quality issues and the expectation of a duty to care for customers. ASUS was founded on engineering principals. Engineering trumps PR. When you have the very best products in the market, then you don’t need to have the it is ‘legal to buy positive reviews’ mentality.  You simply ‘out engineer’ the competition – and ask the most experienced and trusted publications around the world to ‘put your product to the test’. You are already confident you will do well.

The Gamers Nexus call out:

In recent weeks, issues were discovered with Ryzen 7000X3D CPUs and certain motherboards, leading to a wave of beta BIOS versions for various motherboards in an effort to correct the problem. While Asus encouraged users to install this BIOS to avoid issues, they also say that installing the beta BIOS voids your product warranty. Making matters worse, the BIOS in question still ran Ryzen 7000X3D CPUs at unsafe voltages. Despite AMD confirming to Tom's Hardware that the safe voltage for these processors was 1.3V, the Asus BIOS would overvolt the chip to 1.4V anyway, leading to damaged processors and motherboards.

In the Gamers Nexus video above, you will find a detailed breakdown of this situation. Asus had reached out to Gamers Nexus to try and get ahead of the situation, but upon being told that an in-person meeting at the GN offices would need to be recorded, Asus chose to stop sending emails, leaving questions about the company's warranty and support policies unanswered. In all, it seems like Asus is trying to dodge responsibility rather than doing right by its customers, who now regularly pay higher and higher prices for everything from motherboards to graphics cards, cases and coolers.

The Jayz Two Cents video:

In this Jayz Two Cents video, we see that one of the biggest independent tech creators on YouTube, Jayz Two Cents, has also been having problems with Asus recently, to the point where he is now refusing to take them on as a sponsor for any future videos. Jay has experienced a lack of support from Asus PR reps and has noted a number of quality issues with products he has received. For whatever reason, Asus does not appear to be interested in supporting independent media who want to give honest opinions, something that can be further validated by our own experiences with Asus PR over the past year.

___________________________________________

From KitGuru's point of view, we have published thousands of ASUS-related reviews, news stories and posts over the last 13 years. During that time, we have worked with many different people at ASUS – across multiple departments and countries – and the overall feeling was positive. Over the past 12 months, we have felt the situation change and that was brought to a head toward the end of 2022.

Increased pressure to say positive things and public statements from the person responsible for sampling and related budgets, saying that they were “Not expecting any coverage of ASUS at CES 2023, because KitGuru isn’t being paid”.  Frankly, it was disgusting to see and so untrue as to be actionable. But that’s not who we are as a company. No one tells us what to cover. Despite this increasingly negative, non-customer-supporting view being expressed by ASUS, we still covered their 540Hz monitor, the Noctua-enabled RTX 4080 and their latest laptop products. Why? Because those stories are genuinely interesting for you – our reader – and our editorial focus is dictated by readers – not manufacturers.

When the person responsible for ASUS samples and budget is publicly stating that they do not think KitGuru has the capability to ‘review monitors properly’, it makes you question their motivation. That is coming from someone who is OK with ‘buying the results’. We have to assume that they would think KitGuru was ‘better at monitor reviews’ if we took money to say what ASUS wanted to hear. Sorry. We don’t do that and never will.

Our experts test the products, using state of the art tools and proven methods – and the results are what they are. We publish without fear or favour and you cannot buy a result.

Check ASUS UK customer satisfaction rating out on independent website TRUSTPILOT HERE

Kitguru says: We hope that Jonney Shih and Jonathan Tsang look at all of the issues that have surrounded ASUS recently, that they fact-find for themselves and then set their teams back on track. ASUS should focus solely on engineering excellence and customer satisfaction. Secret BIOS settings, passing the blame, removing warranty support and wanting to buy positive reviews – should have no place in the ASUS corporate culture.

ASUS UK PR believes it is ‘legal to buy positive reviews’

This AI girlfriend made $72k in a week

Microsoft News Roundup: Windows 11 'Moment 3,' Pixel Fold vs Surface Duo, ASUS ROG Ally price and spec details, and ASUS motherboard issues

The rise of digital stores like Steam, GoG.com, and more recently the Epic Games Store has contributed to allowing PC games of all types to be sold nearly everywhere. A few weeks ago, we posted our list of the most anticipated Xbox Series X/S games of 2023, so we decided to do the same for PC games as well.

We will mention that all of our games on the Xbox 2023 list will also be released on the PC as well. We didn't want to duplicate our Xbox list for our most anticipated PC game, so we picked some more upcoming titles, many of which will be released either exclusively for PC or first on that platform (although some games may show up on other platforms as well).

System Shock remake - May 30

Developer: Nightdive Studios

Publisher: Prime Matter

What's it about: You have entered a space station controlled by an evil AI, and you have to fight all sorts of enemies in order to defeat it. Good luck with that.

Why we want to play it: Nightdive has indeed taken a long, long time to get this remake finished, but based on what they have promised, it will take the original game's mix of first-person action and role playing to new levels, and we can't wait to check it out

• Pre-order System Shock remake on Steam

   

Amnesia: The Bunker - June 6

Developer: Frictional Games

Publisher: Frictional Games

What's it about: During World War I, you play a French soldier in a bunker with a revolver and a flashlight. That's all you have to face some terrifying horrors.

Why we want to play it: The Amnesia series from Frictional Games have been among the most frightening PC titles ever released, and there's no reason to think this new entry won't be very scary, and very entertaining.

• Pre-order Amnesia: The Bunker on Steam

   

Baldur's Gate III - August 31. 2023

Developer: Larian Studios

Publisher: Larian Studios

What's it about: You have the bad luck of having a parasite implanted into you, which will eventually turn you into a mind flayer. You team up with some other adventurers to defeat this evil, and more importantly, save yourself

Why we want to play it: Taking up from where BioWare left off, Larian Studios is aiming for this to be the ultimate Dungeons and Dragons RPG PC game. It's been in early access for a while but we definitely want to play the full game this year.

• Pre-order Baldur's Gate III on Steam

Counter-Strike 2 - Summer 2023

Developer: Valve

Publisher: Valve

What's it about: You team up with your online friends and shoot other online friends in matches of terrorists and counter-terrorists.

Why we want to play it: It's the long-awaited Source Engine 2 remake of Counter-Strike. Who wouldn't want to play it? And best of all, it's going to be free to play.

Cities: Skylines II - TBD 2023

Developer: Colossal Order

Publisher: Paradox Interactive

What's it about: It's time to create your dream city, with all sorts of cool buildings, highways, and other infrastructure

Why we want to play it: Colossal Order picked up the ball and developed a true spiritual sequel to the Sim City franchise with the original Cities: Skylines. This sequel looks like it will have more of what we liked in that game.

S.T.A.L.K.E.R 2: Heart of Chornobyl - TBD 2023

Developer: GSC GameWorld

Publisher: GSC GameWorld

What's it about: In the near future, the area around the Chornobyl nuclear reactor has been turned into a radiation-filled wasteland filled with mutated creatures and strange occurrences.

Why we want to play it: We have also been waiting a long time for a new S.T.A.L.K.E.R. game. The brief glimpses of the sequel in trailers have given us high hopes it will be worth the over decade wait.

The Finals - TBD 2023

Developer: Embark Studios

Publisher: Embark Studios

What's it about: In a future version of a reality show, you compete with friends in a virtual battleground for money and fame.

Why we want to play it: Embark's recent beta test of this game generated a lot of buzz, mostly because nearly everything on the game's maps could be destroyed. We think the developer will use the beta feedback to give us a multiplayer shooter with some unique twists.

Homeworld 3 - TBD 2023

Developer: Blackbird Interactive

Publisher: Gearbox Publishing

What's it about: A mysterious force is slowly taking over the galaxy and now an advanced civilization is sending their fleet of ships to find this evil and defeat it.

Why we want to play it: Blackbird Interactive has some of the original team members that worked on the first two games in this landmark sci-fi RTS series. Based on the trailers and information, Homeworld 3 could be the ultimate space RTS game.

Hades II - TBD 2023

Developer: Supergiant Games

Publisher: Supergiant Games

What's it about: The Princess of the Underworld is going to try to defeat the Titan of Time with some help from the Olympian goes.

Why we want to play it: The original Hades was a refreshing new take on a dungeon crawler title, and we think the experience from that game will help make the sequel an even deeper and more entertaining experience.

Payday 3 - TBD 2023

Developer: Starbreeze Studios

Publisher: Prime Matter

What's it about: It's time to come out of retirement and start robbing banks again, and take care of anyone who gets in our way.

Why we want to play it: Starbreeze has had its problems, but it appears that the studio is finally going to bring the popular Payday co-op shooter series out of mothballs and into an all-new version, and we can't wait to see it.

That's our list of our most anticipated PC games of 2023. Again, we will be updating this list every so often as more games are announced, and other games get release dates.

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

Here are our picks for the most anticipated PC games of 2023

A few weeks ago, my spouse and I made a bet. I said there was no way ChatGPT could believably mimic my writing style for a smartwatch review. I’d already asked the bot to do that months ago, and the results were laughable. My spouse bet that they could ask ChatGPT the exact same thing but get a much better result. My problem, they said, was I didn’t know the right queries to ask to get the answer I wanted.

 

To my chagrin, they were right. ChatGPT wrote much better reviews as me when my spouse did the asking.

 

That memory flashed through my mind while Iiveblogging Google I/O. This year’s keynote was essentially a two-hour thesis on AI, how it’ll impact Search, and all the ways it could boldly and responsibly make our lives better. A lot of it was neat. But I felt a shiver run down my spine when Google openly acknowledged that it’s hard to ask AI the right questions.

 

During its demo of Duet AI, a series of tools that will live inside Gmail, Docs, and more, Google showed off a feature called Sidekick that can proactively offer you prompts that change based on the Workspace document you’re working on. In other words, it’s prompting you on how to prompt it by telling you what it can do.

 

That showed up again later in the keynote when Google demoed its new AI search results, called Search Generative Experience (SGE). SGE takes any question you type into the search bar and generates a mini report, or a “snapshot,” at the top of the page. At the bottom of that snapshot are follow-up questions.

 

As a person whose job is to ask questions, both demos were unsettling. The queries and prompts Google used on stage look nothing like the questions I type into my search bar. My search queries often read like a toddler talking. (They’re also usually followed by “Reddit” so I get answers from a non-SEO content mill.) Things like “Bald Dennis BlackBerry movie actor name.” When I’m searching for something I wrote about Peloton’s 2022 earnings, I pop in “Site:theverge.com Peloton McCarthy ship metaphors.” Rarely do I search for things like “What should I do in Paris for a weekend?” I don’t even think to ask Google stuff like that.

 

I’ll admit that when staring at any kind of generative AI, I don’t know what I’m supposed to do. I can watch a zillion demos, and still, the blank window taunts me. It’s like I’m back in second grade and my grumpy teacher just called on me for a question I don’t know the answer to. When I do ask something, the results I get are laughably bad — things that would take me more time to make presentable than if I just did it myself.

 

On the other hand, my spouse has taken to AI like a fish to water. After our bet, I watched them play around with ChatGPT for a solid hour. What struck me most was how different our prompts and queries were. Mine were short, open-ended, and broad. My spouse left the AI very little room for interpretation. “You have to hand-hold it,” they said. “You have to feed it exactly everything you need.” Their commands and queries are hyper-specific, long, and often include reference links or data sets. But even they have to rephrase prompts and queries over and over again to get exactly what they’re looking for.

 

The SGE snapshots also prompt you on what to ask it next.

Image: Google

 

This is just ChatGPT. What Google’s pitching goes a step further. Duet AI is meant to pull contextual data from your emails and documents and intuit what you need (which is hilarious since I don’t even know what I need half the time). SGE is designed to answer your questions — even those that don’t have a “right” answer — and then anticipate what you might ask next. For this more intuitive AI to work, programmers have to make it so the AI knows what questions to ask users so that users, in turn, can ask it the right questions. This means that programmers have to know what questions users want answered before they’ve even asked them. It gives me a headache thinking about it.

 

Not to get too philosophical, but you could say all of life is about figuring out the right questions to ask. For me, the most uncomfortable thing about the AI era is I don’t think any of us know what we really want from AI. Google says it’s whatever it showed on stage at I/O. OpenAI thinks it’s chatbots. Microsoft thinks it’s a really horny chatbot. But whenever I talk to the average person about AI these days, the question everybody wants answered is simple. How will AI change and impact my life?

 

The problem is nobody, not even the bots, has a good answer for that yet. And I don’t think we’ll get any satisfactory answer until everyone takes the time to rewire their brains to speak with AI more fluently.

 

 

Google, how do I ask your AI the right questions?

One of the most impressive demos at Google I/O started with a photo of a woman in front of a waterfall. A presenter onstage tapped on the woman, picked her up, and moved her to the other side of the image, with the app automatically filling in the space where she once stood. They then tapped on the overcast sky, and it instantly bloomed into a brighter cloudless blue. In just a matter of seconds, the image had been transformed.

 

The AI-powered tool, dubbed the Magic Editor, certainly lived up to its name during the demo. It’s the kind of tool that Google has been building toward for years. It already has a couple of AI-powered image editing features in its arsenal, including the Magic Eraser, which lets you quickly remove people or objects from the background of an image. But this type of tool takes things up a notch by letting you alter the contents — and potentially, the meaning — of a photo in much more significant ways.

 

The Magic Editor transforms the photo in seconds.

GIF: Google

 

While it’s clear that this tool isn’t flawless — and there remains no firm release date for it — Google’s end goal is clear: to make perfecting photos as easy as just tapping or dragging something on your screen. The company markets the tool as a way to “make complex edits without pro-level editing tools,” allowing you to leverage the power of AI to single out and transform a portion of your photo. That includes the ability to enhance the sky, move and scale subjects, as well as remove parts of an image with just a few taps.

 

Google’s Magic Editor attempts to package all the steps that it would take to make similar edits in a program like Photoshop into a single tap — or, at least, that’s what it looks like from the demo. In Photoshop, for example, you’re stuck using the Content-Aware Move tool (or any of the other methods of your choice) to pick up and move a subject inside of an image. Even then, the photo still might not look quite right, which means you’ll have to pick up other tools, like the Clone Stamp tool or maybe even the Spot Healing Brush, to fix any leftover artifacts or a mismatched background. It’s not the most complicated process ever, but as with most professional creative tools, there’s a definite learning curve for people who are new to the program.

 

I’m all for Google making photo editing tools free and more accessible, given that Photoshop and some of the other image editing apps out there are expensive and pretty unintuitive. But putting powerful and incredibly easy-to-use image editing tools into the hands of, well, just about everyone who downloads Google Photos could transform the way we edit — and look at — photos. There have long been discussions about how far a photo can be edited before it’s no longer a photo, and Google’s tools push us closer to a world where we tap on every image to perfect it, reality or not.

 

Samsung recently brought attention to the power of AI-“enhanced” photos with “Space Zoom,” a feature that’s supposed to let you capture incredible pictures of the Moon on newer Galaxy devices. In March, a Reddit user tried using Space Zoom on an almost unsalvageable image of the Moon and found that Samsung appeared to add craters and other patches that weren’t actually there. Not only does this run the risk of creating a “fake” image of the Moon, but it also leaves actual space photographers in a strange place, as they spend years mastering the art of capturing the night sky, only for the public to often be presented with fakes.

 

A sequence of edits with Google’s Magic Editor.

Image: Google

 

To be fair, there are a ton of similar photography-enhancing features that are built in to smartphone cameras. As my colleague Allison Johnson points out, mobile photography already fakes a lot of things, whether it’s by applying filters or unblurring a photo, and doctored images are nothing new. But Google’s Magic Editor could make a more substantial form of fakery easier and more attractive. In its blog post explaining the tool, Google makes it seem like we’re all in search of perfection, noting that the Magic Editor will provide “more control over the final look and feel of your photo” while getting the chance to fix a missed opportunity that would make a photo look its best.

 

Call me some type of weird photo purist, but I’m not a fan of editing a photo in a way that would alter my memory of an event. If I was taking a picture of a wedding and the sky was cloudy, I wouldn’t think about swapping it for something better. Maybe — just maybe — I might consider moving things around or amping up the sky on a picture I’m posting to social media, but even that seems a little disingenuous. But, again, that’s just me. I could still see plenty of people using the Magic Editor to perfect their photos for social media, which adds to the larger conversation of what exactly we should consider a photo and whether or not that’s something people should be obligated to disclose.

 

Google calls its Magic Editor “experimental technology” that will become available to “select” Pixel phones later this year before rolling out to everyone else. If Google is already adding AI-powered image editing tools to Photos, it seems like it’s only a matter of time before smartphone makers integrate these one-tap tools, like sky replacement or the ability to move a subject, directly into a phone’s camera software. Sometimes, the beauty of a photo is its imperfection. It just seems like smartphone makers are trying to push us farther and farther away from that idea.

 

 

Google’s new Magic Editor pushes us toward AI-perfected fakery

Google’s AI tools embrace the dream of Clippy

Scientists have developed a new method to produce and shape large, high-quality mirrors, significantly thinner than the primary mirrors traditionally employed in space telescopes. These resultant mirrors possess enough flexibility to be rolled up and packed efficiently within a spacecraft during launch.

“Launching and deploying space telescopes is a complicated and costly procedure,” said Sebastian Rabien from Max Planck Institute for Extraterrestrial Physics in Germany. “This new approach — which is very different from typical mirror production and polishing procedures — could help solve weight and packaging issues for telescope mirrors, enabling much larger, and thus more sensitive, telescopes to be placed in orbit.”

The researchers created the mirrors by using chemical vapor deposition to grow membrane mirrors on a rotating liquid inside a vacuum chamber. This allowed them to form parabolic thin membrane that can be used as the primary mirror of a telescope once coated with a reflecting surface such as aluminum. Credit: Sebastian Rabien, Max Planck Institute for Extraterrestrial Physics

 

In the Optica Publishing Group journal Applied Optics, Rabien reports the successful fabrication of parabolic membrane mirror prototypes up to 30 cm in diameter. These mirrors, which could be scaled up to the sizes needed in space telescopes, were created by using chemical vapor deposition to grow membrane mirrors on a rotating liquid inside a vacuum chamber. He also developed a method that uses heat to adaptively correct imperfections that might occur after the mirror is unfolded.

“Although this work only demonstrated the feasibility of the methods, it lays the groundwork for larger packable mirror systems that are less expensive,” said Rabien. “It could make lightweight mirrors that are 15 or 20 meters in diameter a reality, enabling space-based telescopes that are orders of magnitude more sensitive than ones currently deployed or being planned.”

Applying an old process in a new way

The new method was developed during the COVID-19 pandemic, which Rabien says gave him some extra time to think and try out new concepts. “In a long series of tests, we researched many liquids to find out their usability for the process, investigated how the polymer growth can be carried out homogeneously, and worked to optimize the process,” he said.

For chemical vapor deposition, a precursor material is evaporated and thermally split into monomeric molecules. Those molecules deposit on the surfaces in a vacuum chamber and then combine to form a polymer. This process is commonly used to apply coatings such as the ones that make electronics water-resistant, but this is the first time it has been used to create parabolic membrane mirrors with the optical qualities necessary for use in telescopes.

Membrane mirrors made using the new technique are flexible enough to be rolled up. This could be helpful for storing the mirrors inside of a launch vehicle. Credit: Sebastian Rabien, Max Planck Institute for Extraterrestrial Physics

 

To create the precise shape necessary for a telescope mirror, the researchers added a rotating container filled with a small amount of liquid to the inside of the vacuum chamber. The liquid forms a perfect parabolic shape onto which the polymer can grow, forming the mirror base. When the polymer is thick enough, a reflective metal layer is applied to the top via evaporation and the liquid is washed away.

“It has long been known that rotating liquids that are aligned with the local gravitational axis will naturally form a paraboloid surface shape,” said Rabien. “Utilizing this basic physics phenomenon, we deposited a polymer onto this perfect optical surface, which formed a parabolic thin membrane that can be used as the primary mirror of a telescope once coated with a reflecting surface such as aluminum.”

Although other groups have created thin membranes for similar purposes, these mirrors are typically shaped using a high-quality optical mold. Using a liquid to form the shape is much more affordable and can be more easily scaled up to large sizes.

Reshaping a folded mirror

The thin and lightweight mirror created using this technique can easily be folded or rolled up during the trip to space. However, it would be nearly impossible to get it back to the perfect parabolic shape after unpacking. To reshape the membrane mirror, the researchers developed a thermal method that uses a localized temperature change created with light to enable adaptive shape control that can bring the thin membrane into the desired optical shape.

The researchers tested their approach by creating 30-cm diameter membrane mirrors in a vacuum deposition chamber. After much trial and error, they were able to produce high-quality mirrors with a surface shape suitable for telescopes. They also showed that their thermal radiative adaptive shaping method worked well, as demonstrated with an array of radiators and illumination from a digital light projector.

The new membrane-based mirrors could also be used in adaptive optics systems. Adaptive optics can improve the performance of optical systems by using a deformable mirror to compensate for distortion in incoming light. Because the surface of the new membrane mirrors is deformable, these mirrors could be shaped with electrostatic actuators to create deformable mirrors that are less expensive to make than those created with conventional methods.

Next, the researchers plan to apply more sophisticated adaptive control to study how well the final surface can be shaped and how much of an initial distortion can be tolerated. They also plan to create a meter-sized deposition chamber to better study the surface structure and packaging and unfolding processes for a large-scale primary mirror.

Source

Dying Windows 10, momentous Windows 11

The two biggest stories of this week surround Microsoft's latest operating system and its upcoming feature updates. The "Moment 3" update has arrived in the Release Preview Channel, meaning Microsoft is ready to ship it to all customers soon. If you plan to update your computer without waiting for the public release, note that parts of the update are rolling out gradually. Fortunately, we have a guide describing how to get the good stuff right away.

This week, we also published our detailed review of the Windows 11 "Moment 3" update, so be sure to check it. The release is not the biggest we have seen so far, but it contains a bunch of great improvements.

On the negative side of Windows 11, enthusiasts discovered that Microsoft might shove more ads into its operating system. The company is allegedly preparing a new page for the Settings app promoting Microsoft 365, OneDrive, and other services.

Those preferring the good-old Windows 10 also got some news: Microsoft has issued a reminder that Windows 10 version 21H2 is about to reach its end of life. Therefore, the company would soon start force-updating users to version 22H2. Windows 11, on the other hand, remains an optional update for consumers with compatible hardware. Another noteworthy Windows 10-related news is that all Windows 10 20H2 versions are now dead. On May 9, 2023, Microsoft put Enterprise, Education, and IoT SKUs to their final rest.

Updates for everyone!

Alongside ending support for the remaining versions of Windows 10 20H2, Microsoft released scheduled Patch Tuesday updates for supported editions on May 9, 2023:

•     Windows 11 22H2 - 22621.1702 (KB5026372)
•     Windows 11 21H2 - 22000.1936 (KB5026368)
•     Windows 10 - 19042/19044/19045.2965 (KB5026361)

Windows Insiders also received their fair share of new builds to test. KB5026438 arrived in the Beta Channel with fixes for Windows 11, 22000.2001 (the original Windows 11 release) is available for testing in the Release Preview. The same channel offers 19045.3030 for those still using Windows 10. All three updates contain many fixes and minor improvements. Sadly, no Dev or Canary builds this week, so we look forward to what the next one will bring us. Oh, and if you happen to be a fan of Android apps on Windows 11, check out the latest insider update for WSA with the memory reallocation feature, anti-virus software support, and more.

On the gaming side, Microsoft released the May 2023 update Xbox and a few updates for Xbox Insiders, plus a new version of the Xbox app for Windows with a redesigned notification center. A Microsoft executive also suggested that Windows might receive the Quick Resume feature, one of the Xbox Series X and S flagship capabilities.

Finally, the ever-going saga of Microsoft trying to buy Activision-Blizzard is expected to get moving soon with clearance from the European Union. As for the UK's objection to the deal, Activision-Blizzard hired a top lawyer to appeal the recent decision.

Software and security

No week goes by without app updates, new releases, free games, and more. Firefox has released version 113 with a hefty changelog; Google pushed the Mica material to its browser in the Canary Channel (looking much better than what we have in the stable release); and Microsoft shipped Edge 114 in the Beta Channel. On the developer-related news, Microsoft is working on a UI makeover for Visual Studio, the first in over ten years.

Moving to other software, we received a neat tool this week that lets you check out the upcoming "never combine" option in Windows 11. It is one of the most requested taskbar changes for Windows 11, and many insiders were excited to see it coming. Stardock also released Groupy 2, an app that brings tabbed experience to Windows 10 and 11.

We also covered a few security-related stories involving Microsoft and its various products. This week, the company shared a guide for blocking vulnerable Windows boot managers, released an HVCI/memory integrity compatibility check tool for Windows 10 and 11, and fixed the BlackLotus UEFI flaw. The company also got blasted by AV-TEST for the amount of resources Windows' default anti-virus software hogs.

No raises for you!

CNBC reported that Microsoft employees would not receive raises this week. Also, the company would decrease performance bonuses for top executives. These budget-trimming measures seem slightly head-scratching for many in light of positive financial reports and Microsoft leading the way in generative AI development. Microsoft's CEO said the company faces "global macroeconomic uncertainties" so no raises for you this year!

Bing AI

There is no stopping in the stream of updates for AI-powered Bing. Microsoft started rolling out chat history and visual answers for various topics this week. At the same time, the software giant admitted that it is too slow to add new GPUs to keep its AI-based search engine snappy and fast. This might become a problem for the company as it aims to update models for Bing three times a year.

And that is all for this week, folks! As a reminder, you can subscribe to the Neowin Newsletter and receive weekly Microsoft and other news directly to your inbox. If you want more, check out the "Microsoft" tag on our website.

Source

THEY HAVE changed the world by writing software. But techy types are also known for composing lengthy memos in prose, the most famous of which have marked turning points in computing. Think of Bill Gates’s “Internet tidal wave” memo of 1995, which reoriented Microsoft towards the web; or Jeff Bezos’s “API mandate” memo of 2002, which opened up Amazon’s digital infrastructure, paving the way for modern cloud computing.

Now techies are abuzz about another memo, this time leaked from within Google, titled “We have no moat”. Its unknown author details the astonishing progress being made in artificial intelligence (AI)—and challenges some long-held assumptions about the balance of power in this fast-moving industry.

Listen to this story. Enjoy more audio and podcasts on iOS or Android.

AI burst into the public consciousness with the launch in late 2022 of ChatGPT, a chatbot powered by a “large language model” (LLM) made by OpenAI, a startup closely linked to Microsoft. Its success prompted Google and other tech firms to release their own LLM-powered chatbots.

Such systems can generate text and hold realistic conversations because they have been trained using trillions of words taken from the internet.

Training a large LLM takes months and costs tens of millions of dollars. This led to concerns that AI would be dominated by a few deep-pocketed firms.

But that assumption is wrong, says the Google memo. It notes that researchers in the open-source community, using free, online resources, are now achieving results comparable to the biggest proprietary models. It turns out that LLMs can be “fine-tuned” using a technique called low-rank adaptation, or LoRa. This allows an existing LLM to be optimised for a particular task far more quickly and cheaply than training an LLM from scratch.

Activity in open-source AI exploded in March, when LLaMa, a model created by Meta, Facebook’s parent, was leaked online. Although it is smaller than the largest LLMs (its smallest version has 7bn parameters, compared with 540bn for Google’s PaLM) it was quickly fine-tuned to produce results comparable to the original version of ChatGPT on some tasks. As open-source researchers built on each other’s work with LLaMa, “a tremendous outpouring of innovation followed,” the memo’s author writes.

This could have seismic implications for the industry’s future. “The barrier to entry for training and experimentation has dropped from the total output of a major research organisation to one person, an evening, and a beefy laptop,” the Google memo claims. An LLM can now be fine-tuned for $100 in a few hours. With its fast-moving, collaborative and low-cost model, “open-source has some significant advantages that we cannot replicate.” Hence the memo’s title: this may mean Google has no defensive “moat” against open-source competitors. Nor, for that matter, does OpenAI.

Not everyone agrees with this thesis. It is true that the internet runs on open-source software. But people use paid-for, proprietary software, from Adobe Photoshop to Microsoft Windows, as well. AI may find a similar balance. Moreover, benchmarking AI systems is notoriously hard. Yet even if the memo is partly right, the implication is that access to AI technology will be far more democratised than seemed possible even a year ago. Powerful LLMs can be run on a laptop; anyone who wants to can now fine-tune their own AI.

This has both positive and negative implications. On the plus side, it makes monopolistic control of AI by a handful of companies far less likely. It will make access to AI much cheaper, accelerate innovation across the field and make it easier for researchers to analyse the behaviour of AI systems (their access to proprietary models was limited), boosting transparency and safety. But easier access to AI also means bad actors will be able to fine-tune systems for nefarious purposes, such as generating disinformation. It means Western attempts to prevent hostile regimes from gaining access to powerful AI technology will fail. And it makes AI harder to regulate, because the genie is out of the bottle.

Whether Google and its ilk really have lost their moat in AI will soon become apparent. But as with those previous memos, this feels like another turning point for computing.

This article appeared in the Leaders section of the print edition under the headline "A stochastic parrot in every pot?"

Source