AMD has announced plans to cut 4% of its global workforce as it repositions to compete in the AI chip market dominated by Nvidia. The layoffs will affect approximately 1,040 employees of its 26,000-strong workforce reported at the end of 2023. CNBC adds: AMD produces powerful AI accelerators for data centers, including the MI300X, which companies such as Meta and Microsoft purchase as an alternative to Nvidia-based systems. But Nvidia dominates the market for powerful AI chips, with over 80% market share, partially because it developed the core software that AI engineers use to develop programs such as OpenAI's ChatGPT.

Read more of this story at Slashdot.

Apple released Final Cut Pro 11 this week, marking the first major version change in over a decade for its professional video editing software. The update introduces several AI-powered features, including a new "Magnetic Mask" function that automatically tracks objects through video clips for targeted color grading and effects. The suite now offers on-device automatic caption generation for dialogue tracks and adds support for spatial video editing compatible with Apple Vision Pro. Users can adjust the depth of titles and objects for 3D viewing. The update requires macOS 14.6 and at least 8GB of RAM, with some features exclusive to Apple silicon Macs. Existing Final Cut Pro X users will receive the upgrade at no cost, while new users can purchase the software for $299. Accompanying updates include Final Cut Camera for iPhone, which now supports H.265 HEVC format for Apple Log footage on iPhone 15/16 Pro models, and Final Cut Pro for iPad 2.1, featuring enhanced automated color grading tools and new creative assets. Projects created on Mac remain incompatible with the iPad version, PetaPixel reports.

Read more of this story at Slashdot.

An anonymous reader writes: JPL in California announced this week a layoff of 325 workers, about 5% of its workforce, the third major layoff imposed this year. The JPL press release indicates the layoffs are because of NASA budget cutbacks, but does not provide any specificity. The cause centers mostly around NASA's decision to pause its Mars Sample Return project, which JPL was leading. From this report: This is the third round of layoffs at JPL this year, a reduction spurred primarily by major budgetary cuts to the Mars Sample Return mission, which is managed by JPL. NASA directed $310 million this year to the effort to bring Mars rocks back to Earth, a steep drop from the $822.3 million it spent on the program the previous year.

Read more of this story at Slashdot.

Washington Post: The Consumer Financial Protection Bureau has taken steps to place Google under formal federal supervision, an extraordinary move that could subject the technology giant to the regular inspections and other rigorous monitoring that the government imposes on major banks. Google has fiercely resisted the idea over months of highly secretive talks, according to two people familiar with the discussions, who spoke on the condition of anonymity to describe them -- setting up what may ultimately be a major legal clash with vast implications for the CFPB's powers in the digital age. The exact scope of the CFPB's concerns is not clear, and its order does not appear to be final. The political fate of the bureau's work under Director Rohit Chopra is also in doubt, as the watchdog agency braces for potentially significant changes to its leadership and agenda with the return of President-elect Donald Trump to the White House. Formed in the aftermath of the 2008 financial crisis, the CFPB has broad powers to protect consumers from unfair, deceptive or predatory financial practices. That includes the ability to place certain firms under supervision, a status that can afford regulators direct access to the company's internal records to ensure their activities are sound -- and seek fixes if they are not.

Read more of this story at Slashdot.

An anonymous reader quotes a report from TechCrunch: U.K. consumer rights group 'Which?' is filing a legal claim against Apple under competition law on behalf of some 40 million users of iCloud, its cloud storage service. The collective proceeding lawsuit, which is seeking 3 billion pounds in compensation damages (around $3.8 billion at current exchange rates), alleges that Apple has broken competition rules by giving its own cloud storage service preferential treatment and effectively locking people into paying for iCloud at "rip-off" prices. "iOS has a monopoly and is in control of Apple's operating systems and it is incumbent on Apple not to use that dominance to gain an unfair advantage in related markets, like the cloud storage market. But that is exactly what has happened," Which wrote in a press release announcing filing the claim with the U.K.'s Competition Appeal Tribunal (CAT). The lawsuit accuses Apple of encouraging users of its devices to sign up to iCloud for photo storage and other data storage needs, while simultaneously making it difficult for consumers to use alternative storage providers -- including by not allowing them to store or back-up all of their phone's data with a third-party provider. "iOS users then have to pay for the service once photos, notes, messages and other data go over the free 5GB limit," Which noted. The suit also accuses Apple of overcharging U.K. consumers for iCloud subscriptions owing to the lack of competition. "Apple raised the price of iCloud for UK consumers by between 20% and 29% across its storage tiers in 2023," it wrote, saying it's seeking damages for all affected Apple customers -- and estimating that individual consumers could be owed an average of 70 pounds (around $90), depending on how long they've been paying Apple for iCloud services. "Anyone who has 'obtained' iCloud services, including non-paying users, over the nine-year timeframe since the Consumer Rights Act came into force on October 1st, 2015," will be included in the claim. U.K.-based consumers will have to opt-out if they do not want to be included. "Consumers who live outside the U.K. and believe they are eligible to be included must actively opt-in to join the action," adds TechCrunch.

Read more of this story at Slashdot.

The European Union has fined Meta $840 million for unfairly tying its Facebook Marketplace classified ads service to its social network, marking the company's first EU antitrust penalty. The European Commission ruled Meta must stop bundling Marketplace with Facebook's social platform and cease imposing unfair conditions on competing classified ads services. Regulators found Meta exploited Facebook's massive user base to disadvantage rivals and used competitors' advertising data to enhance Marketplace. EU antitrust chief Margrethe Vestager said Meta "tied its online classified ads service Facebook Marketplace to its personal social network Facebook and imposed unfair trading conditions on other online classified ads service providers."

Read more of this story at Slashdot.

An anonymous reader quotes a report from TechCrunch: U.K. consumer rights group 'Which?' is filing a legal claim against Apple under competition law on behalf of some 40 million users of iCloud, its cloud storage service. The collective proceeding lawsuit, which is seeking 3 billion pounds in compensation damages (around $3.8 billion at current exchange rates), alleges that Apple has broken competition rules by giving its own cloud storage service preferential treatment and effectively locking people into paying for iCloud at "rip-off" prices. "iOS has a monopoly and is in control of Apple's operating systems and it is incumbent on Apple not to use that dominance to gain an unfair advantage in related markets, like the cloud storage market. But that is exactly what has happened," Which wrote in a press release announcing filing the claim with the U.K.'s Competition Appeal Tribunal (CAT). The lawsuit accuses Apple of encouraging users of its devices to sign up to iCloud for photo storage and other data storage needs, while simultaneously making it difficult for consumers to use alternative storage providers -- including by not allowing them to store or back-up all of their phone's data with a third-party provider. "iOS users then have to pay for the service once photos, notes, messages and other data go over the free 5GB limit," Which noted. The suit also accuses Apple of overcharging U.K. consumers for iCloud subscriptions owing to the lack of competition. "Apple raised the price of iCloud for UK consumers by between 20% and 29% across its storage tiers in 2023," it wrote, saying it's seeking damages for all affected Apple customers -- and estimating that individual consumers could be owed an average of 70 pounds (around $90), depending on how long they've been paying Apple for iCloud services. "Anyone who has 'obtained' iCloud services, including non-paying users, over the nine-year timeframe since the Consumer Rights Act came into force on October 1st, 2015," will be included in the claim. U.K.-based consumers will have to opt-out if they do not want to be included. "Consumers who live outside the U.K. and believe they are eligible to be included must actively opt-in to join the action," adds TechCrunch.

Read more of this story at Slashdot.

A new study reveals that late work is judged more negatively than on-time submissions, even if delays are minimal or pre-communicated. "The findings suggest that, while you might be tempted to take the maximum allotted time to put the finishing touches to a report, submission or piece of work, the extra effort might not be appreciated by colleagues if it comes at the expense of punctual delivery," reports The Guardian. From the report: The study surveyed thousands of people in the US and UK, including managers, executives, human resources personnel and others whose jobs included an element of evaluating others. Participants were asked to rate pieces of work, such as advertising flyers, art, business proposals, product pitches, photography and news articles. But first, they were told it was either submitted early, on deadline or late. "Late" work was consistently rated as worse in quality than when people were told the same work was completed early or on time. The difference was equivalent to including an objective shortcoming such as not meeting a word count. A missed deadline led evaluators to believe an employee had less integrity, and they reported they would be less willing to work with or assign tasks to that person in the future. "Everyone saw the exact same art contest entry, school submission or business proposal, but they couldn't help but use their knowledge of when it came in to guide their evaluation of how good it was," said Maglio, who co-authored the study with David Fang of Stanford University. Those who eagerly submit work early should be advised that this does not appear to earn a boost in opinion, according to the report in the journal Organizational Behavior and Human Decision Processes. It also didn't matter how late the work was submitted, with one day or one week delays viewed just as negatively -- and that remained the case if the employee gave their manager advance warning. The latest study suggests that it is this inability to plan realistically that is frowned on, with factors beyond an employee's control, such as jury duty, not viewed as negatively. "If the reason why you missed the deadline was beyond your control, you as the employee should let your manager know," said Maglio. "That seems to be one of the few instances in which people cut you a break."

Read more of this story at Slashdot.

During a public joint hearing today titled "Unidentified Anomalous Phenomena: Exposing the Truth," four experts testified that the U.S. is running secret UAP programs, including crash retrieval and reverse-engineering programs for advanced nonhuman technology. Although the Pentagon maintains there's no evidence of alien spacecraft, witnesses like Luis Elizondo and Michael Gold argue that UAPs represent an intelligence enigma and call for open, stigma-free study to address potential security concerns and unknown scientific possibilities. NPR reports: Tim Gallaudet, retired rear admiral, U.S. Navy; CEO of Ocean STL Consulting, LLC "Confirmation that UAPs are interacting with humanity came for me in January 2015," Gallaudet said in his written testimony (PDF). He describes being part of a pre-deployment naval exercise off the U.S. East Coast that culminated in the famous "Go Fast" video, in which a Navy F/A-18 jet's sensors recorded "an unidentified object exhibiting flight and structural characteristics unlike anything in our arsenal." He was among a group of commanders involved in the exercise who received an email containing the video, which was sent by the operations officer of Fleet Forces Command, Gallaudet said. "The very next day, the email disappeared from my account and those of the other recipients without explanation," he said. Luis Elizondo, author and former Department of Defense official Elizondo's written testimony (PDF) was brief and alleged that a secretive arms race is playing out on the global stage. "Let me be clear: UAP are real," he wrote. "Advanced technologies not made by our Government -- or any other government -- are monitoring sensitive military installations around the globe. Furthermore, the U.S. is in possession of UAP technologies, as are some of our adversaries." Elizondo is a former intelligence officer who later "managed a highly sensitive Special Access Program on behalf of the White House and the National Security Council," according to his official bio (PDF). "By 2012, [Elizondo] was the senior ranking person of the DOD's Advanced Aerospace Threat Identification Program, a secretive Pentagon unit that studied unidentified anomalous phenomena," his bio states, adding that he resigned in 2017. Michael Gold, former NASA associate administrator of space policy and partnerships; member of NASA UAP Independent Study Team Gold's written testimony (PDF) stressed the need for government agencies and academics to "overcome the pernicious stigma that continues to impede scientific dialogue and open discussions" about unexplained phenomena. "As the saying goes, the truth is out there," Gold said, "we just need to be bold enough and brave enough to face it." Michael Shellenberger, founder of Public, a news outlet on the Substack platform Shellenberger's testimony (PDF) ran to some 214 pages, including a lengthy timeline of UAP reports from 1947 to 2023. Shellenberger pressed the White House and Congress to act, calling for the adoption of UAP transparency legislation and cutting funds for any related programs that aren't disclosed to lawmakers. "UAP transparency is bi-partisan and critical to our national security," his written testimony stated. You can watch the proceeding here.

Read more of this story at Slashdot.

An anonymous reader quotes a report from Ars Technica: There's a general consensus that we won't be able to consistently perform sophisticated quantum calculations without the development of error-corrected quantum computing, which is unlikely to arrive until the end of the decade. It's still an open question, however, whether we could perform limited but useful calculations at an earlier point. IBM is one of the companies that's betting the answer is yes, and on Wednesday, it announced a series of developments aimed at making that possible. On their own, none of the changes being announced are revolutionary. But collectively, changes across the hardware and software stacks have produced much more efficient and less error-prone operations. The net result is a system that supports the most complicated calculations yet on IBM's hardware, leaving the company optimistic that its users will find some calculations where quantum hardware provides an advantage. [...] Wednesday's announcement was based on the introduction of the second version of its Heron processor, which has 133 qubits. That's still beyond the capability of simulations on classical computers, should it be able to operate with sufficiently low errors. IBM VP Jay Gambetta told Ars that Revision 2 of Heron focused on getting rid of what are called TLS (two-level system) errors. "If you see this sort of defect, which can be a dipole or just some electronic structure that is caught on the surface, that is what we believe is limiting the coherence of our devices," Gambetta said. This happens because the defects can resonate at a frequency that interacts with a nearby qubit, causing the qubit to drop out of the quantum state needed to participate in calculations (called a loss of coherence). By making small adjustments to the frequency that the qubits are operating at, it's possible to avoid these problems. This can be done when the Heron chip is being calibrated before it's opened for general use. Separately, the company has done a rewrite of the software that controls the system during operations. "After learning from the community, seeing how to run larger circuits, [we were able to] almost better define what it should be and rewrite the whole stack towards that," Gambetta said. The result is a dramatic speed-up. "Something that took 122 hours now is down to a couple of hours," he told Ars. Since people are paying for time on this hardware, that's good for customers now. However, it could also pay off in the longer run, as some errors can occur randomly, so less time spent on a calculation can mean fewer errors. Despite all those improvements, errors are still likely during any significant calculations. While it continues to work toward developing error-corrected qubits, IBM is focusing on what it calls error mitigation, which it first detailed last year. [...] The problem here is that using the function is computationally difficult, and the difficulty increases with the qubit count. So, while it's still easier to do error mitigation calculations than simulate the quantum computer's behavior on the same hardware, there's still the risk of it becoming computationally intractable. But IBM has also taken the time to optimize that, too. "They've got algorithmic improvements, and the method that uses tensor methods [now] uses the GPU," Gambetta told Ars. "So I think it's a combination of both."

Read more of this story at Slashdot.

Wi-Fi 8 (also known as IEEE 802.11bn Ultra High Reliability) is expected to arrive around 2028, prioritizing an enhanced user experience over speed by optimizing interactions between devices and access points. While it retains similar bandwidth specifications as the previous standard, Wi-Fi 8 aims to improve network efficiency, reducing interference and congestion for a more reliable and adaptive connection. PCWorld's Mark Hachman reports: As of Nov. 2024, MediaTek believes that Wi-Fi 8 will look virtually identical to Wi-Fi 7 in several key areas: The maximum physical layer (PHY) rate will be the same at 2,880Mbps x 8, or 23Gbits/s. It will also use the same four frequency bands (2, 4, 5, and 6GHz) and the same 4096 QAM modulation across a maximum channel bandwidth of 320MHz. (A Wi-Fi 8 router won't get 23Gbps of bandwidth, of course. According to MediaTek, the actual peak throughput in a "clean," or laboratory, environment is just 80 percent or so of the hypothetical peak throughput, and actual, real-world results can be far less.) Still, put simply, Wi-Fi 8 should deliver the same wireless bandwidth as Wi-Fi 7, using the same channels and the same modulation. Every Wi-Fi standard has also been backwards-compatible with its predecessors, too. What Wi-Fi 8 will do, though, is change how your client device, such as a PC or a phone, interacts with multiple access points. Think of this as an evolution of how your laptop talks to your home's networking equipment. Over time, Wi-Fi has evolved from communications between one laptop and a router, across a single channel. Channel hopping routed different clients to different bands. When Wi-Fi 6 was developed, a dedicated 6GHz channel was added, sometimes as a dedicated "backhaul" between your home's access points. Now, mesh networks are more common, giving your laptop a variety of access points, channels, and frequencies to select between. For a detailed breakdown of the upcoming advancements coming to Wi-Fi 8, including Coordinated Spatial Reuse, Coordinated Beamforming, and Dynamic Sub-Channel Operation, read the full article.

Read more of this story at Slashdot.

The FBI raided Polymarket CEO Shayne Coplan's Manhattan apartment, seizing his phone and electronic devices. A source close to the matter told The New York Post it was politically motivated due to Polymarket's successful prediction of Trump's election win. It's "grand political theater at its worst," the source said. "They could have asked his lawyer for any of these things. Instead, they staged a so-called raid so they can leak it to the media and use it for obvious political reasons." Although no charges were filed, the raid has sparked controversy, with speculation of political retribution and concerns over potential market manipulation, as Polymarket faces scrutiny both in the U.S. and from French regulators. The New York Post reports: Coplan was not arrested and has not been charged, a Polymarket spokesperson told The Post on Wednesday evening. "Polymarket is a fully transparent prediction market that helps everyday people better understand the events that matter most to them, including elections," the rep said. "We charge no fees, take no trading positions, and allow observers from around the world to analyze all market data as a public good." Coplan posted on X after his run-in with the feds: "New phone, who dis?" Polymarket does not allow trading in the US, though bettors can bypass the ban by accessing the site through VPN. The FBI's investigation comes a week after Coplan said Polymarket is planning to return to the US. [...] In 2022, the online gambling platform was forced to pause its trading in the US and pay a $1.4 million penalty to settle charges with the Commodity Futures Trading Commission that it had failed to register with the agency. [In France, regulators are investigating Polymarket's compliance with national gambling laws, with concerns about unauthorized gambling activities within the country.] A Fortune report published a week before the election found widespread evidence of wash-trading on Polymarket. "Polymarket's Terms of Use expressly prohibit market manipulation," a Polymarket spokesperson told Fortune in a statement.

Read more of this story at Slashdot.

Bluesky has reached 15 million users, driven by a recent surge in U.S. signups following the presidential election. It's currently the top free app on iOS. The Verge reports: The platform, which rests on the decentralized AT Protocol, added about a million new users in the last week. Bluesky COO Rose Wang recently told The Verge that the "majority" of new users flocking to the platform have been from the US. Meta's Threads is still outpacing Bluesky, having recently hit 275 million monthly users and growing at a rate of over a million signups per day. But Bluesky offers a very different experience. Both are ad-free (for now), but whereas Threads uses a single Meta-made algorithmic feed, Bluesky offers user-created algorithmic feeds in addition to its "Discover" and "Popular With Friends" ones.

Read more of this story at Slashdot.

An anonymous reader quotes a report from Ars Technica: The classic PC games market is "in a sorry state," according to DRM-free and classic-minded storefront GOG. Small games that aren't currently selling get abandoned, and compatibility issues arise as technology moves forward or as one-off development ideas age like milk. Classic games are only 20 percent of GOG's catalog, and the firm hasn't actually called itself "Good Old Games" in 12 years. And yet, today, GOG announces that it is making "a significant commitment of resources" toward a new GOG Preservation Program. It starts with 100 games for which GOG's own developers are working to create current and future compatibility, keeping them DRM-free and giving them ongoing tech support, along with granting them a "Good Old Game: Preserved by GOG" stamp. GOG is not shifting its mission of providing a DRM-free alternative to Steam, Epic, and other PC storefronts, at least not entirely. But it is demonstrably excited about a new focus that ties back to its original name, inspired in some part by its work on Alpha Protocol. "We think we can significantly impact the classics industry by focusing our resources on it and creating superior products," writes Arthur Dejardin, head of sales and marketing at GOG. "If we wanted to spread the DRM-free gospel by focusing on getting new AAA games on GOG instead, we would make little progress with the same amount of effort and money (we've been trying various versions of that for the last 5 years)." What kind of games? Scanning the list of Good Old Games, most of them are, by all accounts, both good and old. Personally, I'm glad to see the Jagged Alliance games, System Shock 2, Warcraft I & II, Dungeon Keeper Gold and Theme Park, SimCity 3000 Unlimited, and the Wing Commander series (particularly, personally, Privateer). Most of them are, understandably, Windows-only, though Mac support extends to 34 titles so far, and Linux may pick up many more through Proton compatibility, beyond the 19 native titles to date. [...] [I]f you see the shiny foil-ish GOG badge on a game, it's an assurance that GOG has done all it can to bring forward a classic title. It's important work, too. "Preserving" games doesn't just mean locking a stable media in a vault, but keeping games accessible, and playable.

Read more of this story at Slashdot.

Amazon has discontinued its secretive "Encore" project to develop an at-home fertility tracker, resulting in layoffs for around 100 employees. The project, part of Amazon's Grand Challenge division, aimed to launch a device and app that would predict fertility through saliva testing but was ultimately terminated to control costs. CNBC reports: The project was born out of the company's 2020 acquisition of Wisconsin-based startup bluDiagnostics, the sources said. BluDiagnostics was founded in 2015 by Weibel, Katie Brenner and Jodi Schroll, all of whom joined Grand Challenge. The startup had developed a thermometer-like device, called FertilityFinder, to help women track their fertility from home by testing their saliva and measuring two key hormones, estradiol and progesterone. The results of the test were viewable through a corresponding app. Business Insider reported on aspects of the fertility device in 2022, when its codename was Project Tiberius. The team was working to develop its own saliva collection device and mobile app, which could predict when a user might be in the fertile window. Users could also log their period symptoms, sexual activity and other data to assist with tracking their fertility. There are similar offerings on the market from companies including Inne, Oova, Ava and Mira, along with fertility and ovulation tracking apps such as Flo, Clue and Max Levchin's Glow. Amazon initially aimed to release the product this year, but the timing was pushed out after the team encountered technical issues with the device, one of the people said. It was a costly endeavor and required significant upfront investments for lab research and development, in addition to the high salaries for scientists and engineers, the sources said, adding that the team's weekly overhead was roughly $1.5 million. Amazon didn't comment on the figure. Only one project now remains active within Grand Challenge. Its focus is on health tech, the people said. "We regularly review our businesses to ensure we focus on areas where we can make the biggest difference for customers," said Amazon spokesperson Margaret Callahan. "Following a recent review, we've decided to discontinue this project within Grand Challenge, and we're working directly with employees whose roles are impacted to support them through the transition and help them find other opportunities within Amazon."

Read more of this story at Slashdot.

Meta said it plans to introduce advertisements on Threads starting in early 2025, according to a report by The Information (paywalled). GuruFocus reports: Leading the effort -- which is still in its early phases -- is a team inside Instagram's advertising division. One source said Threads is anticipated to let a small number of marketers produce and post material on the platform in January. Threads had about 275 million monthly active users as late as October. During the company's third-quarter earnings call, Meta CEO Mark Zuckerberg observed that Threads daily sign-up count was about one million, each day.

Read more of this story at Slashdot.

An anonymous reader quotes a report from Bloomberg: OpenAI is preparing to launch a new artificial intelligence agent codenamed "Operator" that can use a computer to take actions on a person's behalf (Warning: source may be paywalled; alternative source), such as writing code or booking travel [...]. In a staff meeting on Wednesday, OpenAI's leadership announced plans to release the tool in January as a research preview and through the company's application programming interface for developers [...]. The one nearest completion will be a general-purpose tool that executes tasks in a web browser, one of the people said. OpenAI Chief Executive Officer Sam Altman hinted at the shift to agents in response to a question last month during an Ask Me Anything session on Reddit. "We will have better and better models," Altman wrote. "But I think the thing that will feel like the next giant breakthrough will be agents." The move to release an agentic AI tool also comes as OpenAI and its competitors have seen diminishing returns from their costly efforts to develop more advanced AI models.

Read more of this story at Slashdot.

Microsoft's gaming division is developing prototypes for a handheld gaming device that won't launch for "a few years," gaming chief Phil Spencer said Wednesday. In an interview with Bloomberg, Spencer said that while Microsoft is actively working on prototypes, the company will first focus on improving its Xbox app performance on existing portable devices and establishing hardware partnerships. The gaming unit wants to be "informed by learning and what's happening now" before introducing its own device, Spencer said. "Longer term, I love us building devices," Spencer said, adding that Microsoft's team "could do some real innovative work."

Read more of this story at Slashdot.

Italy has emerged as a major global spyware hub alongside Israel and India, with at least six major vendors operating in the country with limited oversight, The Record reported this week, citing researchers and Italian experts. Companies like RCS Labs, which has operated since 1992, sell surveillance tools to both domestic law enforcement and foreign governments including Kazakhstan, Syria, and several Asian nations. Italian authorities can rent spyware for $160 per day without large acquisition costs, leading to thousands of domestic surveillance operations in recent years. While new regulations taking effect in February 2024 will require judges to evaluate specific reasons for spyware use, critics cited in the story say the reform package won't address core issues like the lack of centralized oversight. The country's competitive marketplace and relatively lax export controls have also enabled Italian vendors to expand their overseas sales.

Read more of this story at Slashdot.

Leading AI systems are solving less than 2% of problems in a new advanced mathematics benchmark, revealing significant limitations in their reasoning capabilities, research group Epoch AI reported this week. The benchmark, called FrontierMath, consists of hundreds of original research-level mathematics problems developed in collaboration with over 60 mathematicians, including Fields Medalists Terence Tao and Timothy Gowers. While top AI models like GPT-4 and Gemini 1.5 Pro achieve over 90% accuracy on traditional math tests, they struggle with FrontierMath's problems, which span computational number theory to algebraic geometry and require complex reasoning. "These are extremely challenging. [...] The only way to solve them is by a combination of a semi-expert like a graduate student in a related field, maybe paired with some combination of a modern AI and lots of other algebra packages," Tao said. The problems are designed to be "guessproof," with large numerical answers or complex mathematical objects as solutions, making it nearly impossible to solve without proper mathematical reasoning. Further reading: New secret math benchmark stumps AI models and PhDs alike.

Read more of this story at Slashdot.