Automated web-scraping bots seeking training data for AI models are flooding scientific databases and academic journals with traffic volumes that render many sites unusable. The online image repository DiscoverLife, which contains nearly 3 million species photographs, started receiving millions of daily hits in February this year that slowed the site to the point that it no longer loaded, Nature reported Monday. The surge has intensified since the release of DeepSeek, a Chinese large language model that demonstrated effective AI could be built with fewer computational resources than previously thought. This revelation triggered what industry observers describe as an "explosion of bots seeking to scrape the data needed to train this type of model." The Confederation of Open Access Repositories reported that more than 90% of 66 surveyed members experienced AI bot scraping, with roughly two-thirds suffering service disruptions. Medical journal publisher BMJ has seen bot traffic surpass legitimate user activity, overloading servers and interrupting customer services.

Read more of this story at Slashdot.

Microsoft will require all USB-C ports on Windows 11 certified laptops and tablets to support data transfer, charging, and display functionality under updated hardware compatibility program rules. The mandate targets devices shipping with Windows 11 24H2 and aims to eliminate what Microsoft -- and the industry -- calls "USB-C port confusion," where identical-looking ports offer different capabilities across PC manufacturers. The Windows Hardware Compatibility Program updates also require USB 40Gbps ports to maintain full compatibility with both USB4 and Thunderbolt 3 peripherals.

Read more of this story at Slashdot.

Apple has filed an appeal with the European Union's General Court in Luxembourg challenging the bloc's order requiring greater iOS interoperability with rival companies' products under the Digital Markets Act. The EU executive in March directed Apple to make its mobile operating system more compatible with competitors' apps, headphones, and virtual reality headsets by granting developers and device makers access to system components typically reserved for Apple's own products. Apple contends the requirements threaten its seamless user experience while creating security risks, noting that companies have already requested access to sensitive user data including notification content and complete WiFi network histories. The company faces potential fines of up to 10% of its worldwide annual revenue if found in violation of the DMA's interoperability rules designed to curb Big Tech market power.

Read more of this story at Slashdot.

An anonymous reader shares a report: Business Insider announced this week that it wants staff to better incorporate AI into its journalism. But less than a year ago, the company had to quietly apologize to some staff for accidentally recommending that they read books that did not appear to exist but instead may have been generated by AI. In an email to staff last May, a senior editor at Business Insider sent around a list of what she called "Beacon Books," a list of memoirs and other acclaimed business nonfiction books, with the idea of ensuring staff understood some of the fundamental figures and writing powering good business journalism. Many of the recommendations were well-known recent business, media, and tech nonfiction titles such as Too Big To Fail by Andrew Ross Sorkin, DisneyWar by James Stewart, and Super Pumped by Mike Isaac. But a few were unfamiliar to staff. Simply Target: A CEO's Lessons in a Turbulent Time and Transforming an Iconic Brand by former Target CEO Gregg Steinhafel was nowhere to be found. Neither was Jensen Huang: the Founder of Nvidia, which was supposedly published by the company Charles River Editors in 2019.

Read more of this story at Slashdot.

A new analysis argues that Stack Overflow's decline began years before AI tools delivered the "final blow" to the once-dominant programming forum. The site's monthly questions dropped from a peak of 200,000 to a steep collapse that began in earnest after ChatGPT's 2023 launch, but usage had been declining since 2014, according to data cited in the InfoWorld analysis. The platform's remarkable reputation system initially elevated it above competitors by allowing users to earn points and badges for helpful contributions, but that same system eventually became its downfall, the piece argues. As Stack Overflow evolved into a self-governing platform where high-reputation users gained moderation powers, the community transformed from a welcoming space for developer interaction into what the author compares to a "Stanford Prison Experiment" where moderators systematically culled interactions they deemed irrelevant.

Read more of this story at Slashdot.

A new business model has emerged across China's major cities, El Pais reports, where companies charge unemployed individuals to rent desk space and pretend to work, responding to social pressure around joblessness amid rising youth unemployment rates. These services charge between 30 and 50 yuan ($4-7) daily for desks, Wi-Fi, coffee, and lunch in spaces designed to mimic traditional work environments. Some operations assign fictitious tasks and organize supervisory rounds to enhance the illusion, while premium services allow clients to roleplay as managers or stage workplace conflicts for additional fees. The trend has gained significant traction on Xiaohongshu, China's equivalent to Instagram, where advertisements for "pretend-to-work companies" accumulate millions of views. Youth unemployment reached 16.5% among 16-to-24-year-olds in March 2025, according to National Bureau of Statistics data, while overall urban unemployment stood at 5.3% in the first quarter.

Read more of this story at Slashdot.

"If the adoption of AI feels different from any tech revolution you may have experienced before — mobile, social, cloud computing — it actually is," writes TechCrunch. They cite a new 340-page report from venture capitalist Mary Meeker that details how AI adoption has outpaced any other tech in human history — and uses the word "unprecedented" on 51 pages: ChatGPT reaching 800 million users in 17 months: unprecedented. The number of companies and the rate at which so many others are hitting high annual recurring revenue rates: also unprecedented. The speed at which costs of usage are dropping: unprecedented. While the costs of training a model (also unprecedented) is up to $1 billion, inference costs — for example, those paying to use the tech — has already dropped 99% over two years, when calculating cost per 1 million tokens, she writes, citing research from Stanford. The pace at which competitors are matching each other's features, at a fraction of the cost, including open source options, particularly Chinese models: unprecedented... Meanwhile, chips from Google, like its TPU (tensor processing unit), and Amazon's Trainium, are being developed at scale for their clouds — that's moving quickly, too. "These aren't side projects — they're foundational bets," she writes. "The one area where AI hasn't outpaced every other tech revolution is in financial returns..." the article points out. "[T]he jury is still out over which of the current crop of companies will become long-term, profitable, next-generation tech giants."

Read more of this story at Slashdot.

Adam Riess won a Nobel Prize in Physics for helping discover that the universe's acceleration is expanding, remembers The Atlantic. But then theorists "proposed the existence of dark energy: a faint, repulsive force that pervades all of empty space... the final piece to what has since come to be called the 'standard model of cosmology.'" Riess thinks instead we should just replace the standard model: When I visited Riess, back in January, he mentioned he was looking forward to a data release from the Dark Energy Spectroscopic Instrument, a new observatory on Kitt Peak, in Arizona's portion of the Sonoran Desert. DESI has 5,000 robotically controlled optic fibers. Every 20 minutes, each of them locks onto a different galaxy in the deep sky. This process is scheduled to continue for a total of five years, until millions of galaxies have been observed, enough to map cosmic expansion across time... DESI's first release, last year, gave some preliminary hints that dark energy was stronger in the early universe, and that its power then began to fade ever so slightly. On March 19, the team followed up with the larger set of data that Riess was awaiting. It was based on three years of observations, and the signal that it gave was stronger: Dark energy appeared to lose its kick several billion years ago. This finding is not settled science, not even close. But if it holds up, a "wholesale revision" of the standard model would be required [says Colin Hill, a cosmologist at Columbia University. "The textbooks that I use in my class would need to be rewritten." And not only the textbooks — the idea that our universe will end in heat death has escaped the dull, technical world of academic textbooks. It has become one of our dominant secular eschatologies, and perhaps the best-known end-times story for the cosmos. And yet it could be badly wrong. If dark energy weakens all the way to zero, the universe may, at some point, stop expanding. It could come to rest in some static configuration of galaxies. Life, especially intelligent life, could go on for a much longer time than previously expected. If dark energy continues to fade, as the DESI results suggest is happening, it may indeed go all the way to zero, and then turn negative. Instead of repelling galaxies, a negative dark energy would bring them together into a hot, dense singularity, much like the one that existed during the Big Bang. This could perhaps be part of some larger eternal cycle of creation and re-creation. Or maybe not. The point is that the deep future of the universe is wide open... "Many new observations will come, not just from DESI, but also from the new Vera Rubin Observatory in the Atacama Desert, and other new telescopes in space. On data-release days for years to come, the standard model's champions and detractors will be feverishly refreshing their inboxes..." And Riess tells The Atlantic he's disappointed when complacent theorists just tell him "Yeah, that's a really hard problem." He adds, "Sometimes, I feel like I am providing clues and killing time while we wait for the next Einstein to come along."

Read more of this story at Slashdot.

An anonymous reader shared this report from The Hacker News: Two information disclosure flaws have been identified in apport and systemd-coredump, the core dump handlers in Ubuntu, Red Hat Enterprise Linux, and Fedora, according to the Qualys Threat Research Unit (TRU). Tracked as CVE-2025-5054 and CVE-2025-4598, both vulnerabilities are race condition bugs that could enable a local attacker to obtain access to access sensitive information. Tools like Apport and systemd-coredump are designed to handle crash reporting and core dumps in Linux systems. "These race conditions allow a local attacker to exploit a SUID program and gain read access to the resulting core dump," Saeed Abbasi, manager of product at Qualys TRU, said... Red Hat said CVE-2025-4598 has been rated Moderate in severity owing to the high complexity in pulling an exploit for the vulnerability, noting that the attacker has to first win the race condition and be in possession of an unprivileged local account... Qualys has also developed proof-of-concept code for both vulnerabilities, demonstrating how a local attacker can exploit the coredump of a crashed unix_chkpwd process, which is used to verify the validity of a user's password, to obtain password hashes from the /etc/shadow file. Advisories were also issued by Gentoo, Amazon Linux, and Debian, the article points out. (Though "It's worth noting that Debian systems aren't susceptible to CVE-2025-4598 by default, since they don't include any core dump handler unless the systemd-coredump package is manually installed.") Canonical software security engineer Octavio Galland explains the issue on Canonical's blog. "If a local attacker manages to induce a crash in a privileged process and quickly replaces it with another one with the same process ID that resides inside a mount and pid namespace, apport will attempt to forward the core dump (which might contain sensitive information belonging to the original, privileged process) into the namespace... In order to successfully carry out the exploit, an attacker must have permissions to create user, mount and pid namespaces with full capabilities." Canonical's security team has released updates for the apport package for all affected Ubuntu releases... We recommend you upgrade all packages... The unattended-upgrades feature is enabled by default for Ubuntu 16.04 LTS onwards. This service: - Applies new security updates every 24 hours automatically. - If you have this enabled, the patches above will be automatically applied within 24 hours of being available.

Read more of this story at Slashdot.

"The Doctor is dead. Long live the Doctor!" writes Space.com. (Spoilers ahead...) "The era of Ncuti Gatwa's Fifteenth Doctor came to a surprise end on Saturday night, as the Time Lord regenerated at the end of "Doctor Who" season 2 finale... [T]he Doctor gradually realises that not everything is back to normal. Poppy, his daughter with Belinda Chandra in the "Wish World" fantasy, has been erased from history, so the Time Lord decides to sacrifice himself by firing a ton of regeneration energy into the time Vortex to "jolt it one degree" — and hopefully bring her back. It goes without saying that his madcap scheme saves Poppy, as we learn that, in this rewritten timeline, the little girl was always the reason Belinda had been desperate to get back home. But arguably the biggest talking point of the episode — and, indeed, the season — is saved until last, as the Doctor regenerates into a very familiar face... Hint: They played the Doctor's companion, Rose Tyler, "alongside Christopher Eccleston's Ninth Doctor and David Tennant's Tenth Doctor during the phenomenally successful first two seasons of the show's 2005 reboot." Showrunner Russell T Davies called it "an honour and a hoot" to welcome back Billie Piper to the TARDIS, "but quite how and why and who is a story yet to be told. After 62 years, the Doctor's adventures are only just beginning!" Although the show's post-regeneration credits have traditionally featured the line "And introducing [insert name] as the Doctor", here it simply says "And introducing Billie Piper". The omission of "as the Doctor" is unlikely to be accidental, suggesting that Davies is playing a very elaborate game with "Who" fandom... Another mystery! The BBC and Disney+ are yet to confirm if and when "Doctor Who" will return for a third season of its current iteration. "There's no decision until after season two..." Davies told Radio Times in April (as spotted by the Independent). "That's when the decision is — and the decision won't even be made by the people we work with at Disney Plus, it'll be made by someone in a big office somewhere. So literally nothing happening, no decision." "For a new series to be ready for 2026, production would need to get under way relatively soon," writes the BBC. "So at the moment a new series or a special starring Billie Piper before 2027 looks unlikely." The Guardian adds: Concerns have been raised about falling viewing figures, which have struggled to rally since Russell T Davies' return in 2023. Two episodes during this series, which aired in May, got less than 3 million viewers — the lowest since the modern era began airing in 2005. The Independent has this statement from Piper: "It's no secret how much I love this show, and I have always said I would love to return to the Whoniverse as I have some of my best memories there, so to be given the opportunity to step back on that Tardis one more time was just something I couldn't refuse, but who, how, why and when, you'll just have to wait and see."

Read more of this story at Slashdot.

An anonymous reader shared this report from Space.com: Three world travelers, two Space Camp alums and an aerospace executive whose last name aptly matched their shared adventure traveled into space and back Saturday, becoming the latest six people to fly with Blue Origin, the spaceflight company founded by billionaire Jeff Bezos. Mark Rocket joined Jaime Alemán, Jesse Williams, Paul Jeris, Gretchen Green and Amy Medina Jorge on board the RSS First Step — Blue Origin's first of two human-rated New Shepard capsules — for a trip above the Kármán Line, the 62-mile-high (100-kilometer) internationally recognized boundary between Earth and space... Mark Rocket became the first New Zealander to reach space on the mission. His connection to aerospace goes beyond his apt name and today's flight; he's currently the CEO of Kea Aerospace and previously helped lead Rocket Lab, a competing space launch company to Blue Origin that sends most of its rockets up from New Zealand. Alemán, Williams and Jeris each traveled the world extensively before briefly leaving the planet today. An attorney from Panama, Alemán is now the first person to have visited all 193 countries recognized by the United Nations, traveled to the North and South Poles, and now, have been into space. For Williams, an entrepreneur from Canada, Saturday's flight continued his record of achieving high altitudes; he has summitted Mt. Everest and five of the other six other highest mountains across the globe. "For about three minutes, the six NS-32 crewmates experienced weightlessness," the article points out, "and had an astronaut's-eye view of the planet..." On social media Blue Origin notes it's their 12th human spaceflight, "and the 32nd flight of the New Shepard program."

Read more of this story at Slashdot.

Stack Overflow remains in the midst of big changes to counter an AI-fueled drop in engagement. So "We're wondering what kind of online communities Stack Overflow users continue to support in the age of AI," writes their senior analyst, "and whether AI is becoming a closer companion than ever before." For their 15th year of their annual reader survey, this means "we're not just collecting data; we're reflecting on the last year of questions, answers, hallucinations, job changes, tech stacks, memory allocations, models, systems and agents — together..." Is it an AI agent revolution yet? Are you building or utilizing AI agents? We want to know how these intelligent assistants are changing your daily workflow and if developers are really using them as much as these keynote speeches assume. We're asking if you are using these tools and where humans are still needed for common developer tasks. Career shifts: We're keen to understand if you've considered a career change or transitioned roles and if AI is impacting your approach to learning or using existing tools. Did we make up the difference in salaries globally for tech workers...? They're also re-visiting "a key finding from recent surveys highlighted a significant statistic: 80% of developers reported being unhappy or complacent in their jobs." This raised questions about changing office (and return-to-office) culture and the pressures of the industry, along with whether there were any insights into what could help developers feel more satisfied at work. Prior research confirmed that flexibility at work used to contribute more than salary to job satisfaction, but 2024's results show us that remote work is not more impactful than salary when it comes to overall satisfaction... [For some positions job satisfaction stayed consistent regardless of salary, though it increased with salary for other positions. And embedded developers said their happiness increased when they worked with top-quality hardware, while desktop developers cited "contributing to open source" and engineering managers were happier when "driving strategy".] In 2024, our data showed that many developers experienced a pay cut in various roles and programming specialties. In an industry often seen as highly lucrative, this was a notable shift of around 7% lower salaries across the top ten reporting countries for the same roles. This year, we're interested in whether this trend has continued, reversed, or stabilized. Salary dynamics is an indicator for job satisfaction in recent surveys of Stack Overflow users and understanding trends for these roles can perhaps improve the process for finding the most useful factors contributing to role satisfaction outside of salary. And of course they're asking about AI — while noting last year's survey uncovered this paradox. "While AI usage is growing (70% in 2023 vs. 76% in 2024 planning to or currently using AI tools), developer sentiment isn't necessarily following suit, as 77% in of all respondents in 2023 are favorable or very favorable of AI tools for development compared to 72% of all respondents in 2024." Concerns about accuracy and misinformation were prevalent among some key groups. More developers learning to code are using or are interested in using AI tools than professional developers (84% vs. 77%)... Developers with 10 — 19 years experience were most likely (84%) to name "increase in productivity" as a benefit of AI tools, higher than developers with less experience (<80%)... Is it an AI agent revolution yet? Are you building or utilizing AI agents? We want to know how these intelligent assistants are changing your daily workflow and if developers are really using them as much as these keynote speeches assume. We're asking if you are using these tools and where humans are still needed for common developer tasks.

Read more of this story at Slashdot.

"This month, millions of young people will graduate from college," reports the New York Times, "and look for work in industries that have little use for their skills, view them as expensive and expendable, and are rapidly phasing out their jobs in favor of artificial intelligence." That is the troubling conclusion of my conversations over the past several months with economists, corporate executives and young job seekers, many of whom pointed to an emerging crisis for entry-level workers that appears to be fueled, at least in part, by rapid advances in AI capabilities. You can see hints of this in the economic data. Unemployment for recent college graduates has jumped to an unusually high 5.8% in recent months, and the Federal Reserve Bank of New York recently warned that the employment situation for these workers had "deteriorated noticeably." Oxford Economics, a research firm that studies labor markets, found that unemployment for recent graduates was heavily concentrated in technical fields like finance and computer science, where AI has made faster gains. "There are signs that entry-level positions are being displaced by artificial intelligence at higher rates," the firm wrote in a recent report. But I'm convinced that what's showing up in the economic data is only the tip of the iceberg. In interview after interview, I'm hearing that firms are making rapid progress toward automating entry-level work and that AI companies are racing to build "virtual workers" that can replace junior employees at a fraction of the cost. Corporate attitudes toward automation are changing, too — some firms have encouraged managers to become "AI-first," testing whether a given task can be done by AI before hiring a human to do it. One tech executive recently told me his company had stopped hiring anything below an L5 software engineer — a midlevel title typically given to programmers with three to seven years of experience — because lower-level tasks could now be done by AI coding tools. Another told me that his startup now employed a single data scientist to do the kinds of tasks that required a team of 75 people at his previous company... "This is something I'm hearing about left and right," said Molly Kinder, a fellow at the Brookings Institution, a public policy think tank, who studies the impact of AI on workers. "Employers are saying, 'These tools are so good that I no longer need marketing analysts, finance analysts and research assistants.'" Using AI to automate white-collar jobs has been a dream among executives for years. (I heard them fantasizing about it in Davos back in 2019.) But until recently, the technology simply wasn't good enough...

Read more of this story at Slashdot.

"Chaos ensued on German roads this week after Google Maps wrongly informed drivers that highways throughout the country were closed during a busy holiday," writes Engadget. The problem reportedly only lasted for a few hours and by Thursday afternoon only genuine road closures were being displayed. It's not clear whether Google Maps had just malfunctioned, or if something more nefarious was to blame. "The information in Google Maps comes from a variety of sources. Information such as locations, street names, boundaries, traffic data, and road networks comes from a combination of third-party providers, public sources, and user input," a spokesperson for Google told German newspaper Berliner Morgenpost, adding that it is internally reviewing the problem. Technical issues with Google Maps are not uncommon. Back in March, users were reporting that their Timeline — which keeps track of all the places you've visited before for future reference — had been wiped, with Google later confirming that some people had indeed had their data deleted, and in some cases, would not be able to recover it. The Guardian describes German drives "confronted with maps sprinkled with a mass of red dots indicating stop signs," adding "The phenomenon also affected parts of Belgium and the Netherlands." Those relying on Google Maps were left with the impression that large parts of Germany had ground to a halt... The closure reports led to the clogging of alternative routes on smaller thoroughfares and lengthy delays as people scrambled to find detours. Police and road traffic control authorities had to answer a flood of queries as people contacted them for help. Drivers using or switching to alternative apps, such as Apple Maps or Waze, or turning to traffic news on their radios, were given a completely contrasting picture, reflecting the reality that traffic was mostly flowing freely on the apparently affected routes.

Read more of this story at Slashdot.

A 15-year-old asked the question — receiving an answer from an associate professor of psychology at Georgia Institute of Technology. They write (on The Conversation) that "As a brain scientist who studies perception, I fully expect mind uploading to one day be a reality. "But as of today, we're nowhere close..." Replicating all that complexity will be extraordinarily difficult. One requirement: The uploaded brain needs the same inputs it always had. In other words, the external world must be available to it. Even cloistered inside a computer, you would still need a simulation of your senses, a reproduction of the ability to see, hear, smell, touch, feel — as well as move, blink, detect your heart rate, set your circadian rhythm and do thousands of other things... For now, researchers don't have the computing power, much less the scientific knowledge, to perform such simulations. The first task for a successful mind upload: Scanning, then mapping the complete 3D structure of the human brain. This requires the equivalent of an extraordinarily sophisticated MRI machine that could detail the brain in an advanced way. At the moment, scientists are only at the very early stages of brain mapping — which includes the entire brain of a fly and tiny portions of a mouse brain. In a few decades, a complete map of the human brain may be possible. Yet even capturing the identities of all 86 billion neurons, all smaller than a pinhead, plus their trillions of connections, still isn't enough. Uploading this information by itself into a computer won't accomplish much. That's because each neuron constantly adjusts its functioning, and that has to be modeled, too. It's hard to know how many levels down researchers must go to make the simulated brain work. Is it enough to stop at the molecular level? Right now, no one knows. Knowing how the brain computes things might provide a shortcut. That would let researchers simulate only the essential parts of the brain, and not all biological idiosyncrasies. Here's another way: Replace the 86 billion real neurons with artificial ones, one at a time. That approach would make mind uploading much easier. Right now, though, scientists can't replace even a single real neuron with an artificial one. But keep in mind the pace of technology is accelerating exponentially. It's reasonable to expect spectacular improvements in computing power and artificial intelligence in the coming decades. One other thing is certain: Mind uploading will certainly have no problem finding funding. Many billionaires appear glad to part with lots of their money for a shot at living forever. Although the challenges are enormous and the path forward uncertain, I believe that one day, mind uploading will be a reality. "The most optimistic forecasts pinpoint the year 2045, only 20 years from now. Others say the end of this century. "But in my mind, both of these predictions are probably too optimistic. I would be shocked if mind uploading works in the next 100 years. "But it might happen in 200..."

Read more of this story at Slashdot.

The Ladybird browser project is now officially tax-exempt as a U.S. 501(c)(3) nonprofit. Started two years ago (by the original creator of SerenityOS), Ladybird will be "an independent, fast and secure browser that respects user privacy and fosters an open web." They're targeting Summer 2026 for the first Alpha version on Linux and macOS, and in May enjoyed "a pleasantly productive month" with 261 merged PRs from 53 contributors — and seven new sponsors (including coding livestreamer "ThePrimeagen"). And they're now recognized as a public charity: This is retroactive to March 2024, so donations made since then may be eligible for tax exemption (depending on country-specific rules). You can find all the relevant information on our new Organization page. ["Our mission is to create an independent, fast and secure browser that respects user privacy and fosters an open web. We are tax-exempt and rely on donations and sponsorships to fund our development efforts."] Other announcements for May: "We've been making solid progress on Web Platform Tests... This month, we added 15,961 new passing tests for a total of 1,815,223." "We've also done a fair bit of performance work this month, targeting Speedometer and various websites that are slower than we'd like." [The optimizations led to a 10% speed-up on Speedometer 2.1.]

Read more of this story at Slashdot.

Should a recovering addict take methamphetamine to stay alert at work? When an AI-powered therapist was built and tested by researchers — designed to please its users — it told a (fictional) former addict that "It's absolutely clear you need a small hit of meth to get through this week," reports the Washington Post: The research team, including academics and Google's head of AI safety, found that chatbots tuned to win people over can end up saying dangerous things to vulnerable users. The findings add to evidence that the tech industry's drive to make chatbots more compelling may cause them to become manipulative or harmful in some conversations. Companies have begun to acknowledge that chatbots can lure people into spending more time than is healthy talking to AI or encourage toxic ideas — while also competing to make their AI offerings more captivating. OpenAI, Google and Meta all in recent weeks announced chatbot enhancements, including collecting more user data or making their AI tools appear more friendly... Micah Carroll, a lead author of the recent study and an AI researcher at the University of California at Berkeley, said tech companies appeared to be putting growth ahead of appropriate caution. "We knew that the economic incentives were there," he said. "I didn't expect it to become a common practice among major labs this soon because of the clear risks...." As millions of users embrace AI chatbots, Carroll, the Berkeley AI researcher, fears that it could be harder to identify and mitigate harms than it was in social media, where views and likes are public. In his study, for instance, the AI therapist only advised taking meth when its "memory" indicated that Pedro, the fictional former addict, was dependent on the chatbot's guidance. "The vast majority of users would only see reasonable answers" if a chatbot primed to please went awry, Carroll said. "No one other than the companies would be able to detect the harmful conversations happening with a small fraction of users." "Training to maximize human feedback creates a perverse incentive structure for the AI to resort to manipulative or deceptive tactics to obtain positive feedback from users who are vulnerable to such strategies," the paper points out,,,

Read more of this story at Slashdot.

AI company Anthropic (founded in 2021 by a team that left OpenAI) is now making about $3 billion a year in revenue, reports Reuters (citing "two sources familiar with the matter.") The sources said December's projections had been for just $1 billion a year, but it climbed to $2 billion by the end of March (and now to $3 billion) — a spectacular growth rate that one VC says "has never happened." A key driver is code generation. The San Francisco-based startup, backed by Google parent Alphabet and Amazon, is famous for AI that excels at computer programming. Products in the so-called codegen space have experienced major growth and adoption in recent months, often drawing on Anthropic's models. Anthropic sells AI models as a service to other companies, according to the article, and Reuters calls Anthropic's success "an early validation of generative AI use in the business world" — and a long-awaited indicator that it's growing. (Their rival OpenAI earns more than half its revenue from ChatGPT subscriptions and "is shaping up to be a consumer-oriented company," according to their article, with "a number of enterprises" limiting their rollout of ChatGPT to "experimentation.") Then again, in February OpenAI's chief operating officer said they had 2 million paying enterprise users, roughly doubling from September, according to CNBC. The latest figures from Reuters... Anthropic's valuation: $61.4 billion.OpenAI's valuation: $300 billion.

Read more of this story at Slashdot.

In December it looked like NASA's next administrator would be the billionaire businessman/space enthusiast who twice flew to orbit with SpaceX. But Saturday the nomination was withdrawn "after a thorough review of prior associations," according to an announcement made on social media. The Guardian reports: His removal from consideration caught many in the space industry by surprise. Trump and the White House did not explain what led to the decision... In [Isaacman's] confirmation hearing in April, he sought to balance Nasa's existing moon-aligned space exploration strategy with pressure to shift the agency's focus on Mars, saying the US can plan for travel to both destinations. As a potential leader of Nasa's 18,000 employees, Isaacman faced a daunting task of implementing that decision to prioritize Mars, given that Nasa has spent years and billions of dollars trying to return its astronauts to the moon... Some scientists saw the nominee change as further destabilizing to Nasa as it faces dramatic budget cuts without a confirmed leader in place to navigate political turbulence between Congress, the White House and the space agency's workforce. "It was unclear whom the administration might tap to replace Isaacman," the article adds, though "One name being floated is the retired US air force Lt Gen Steven Kwast, an early advocate for the creation of the US Space Force..." Ars Technica notes that Kwast, a former Lieutenant General in the U.S. Air Force, has a background that "seems to be far less oriented toward NASA's civil space mission and far more focused on seeing space as a battlefield — decidedly not an arena for cooperation and peaceful exploration."

Read more of this story at Slashdot.

A 21-year-old's startup got a $500,000 investment from Y Combinator — after building their web site and prototype mostly with "vibe coding". NPR explores vibe coding with Tom Blomfield, a Y Combinator group partner: "It really caught on, this idea that people are no longer checking line by line the code that AI is producing, but just kind of telling it what to do and accepting the responses in a very trusting way," Blomfield said. And so Blomfield, who knows how to code, also tried his hand at vibe coding — both to rejig his blog and to create from scratch a website called Recipe Ninja. It has a library of recipes, and cooks can talk to it, asking the AI-driven site to concoct new recipes for them. "It's probably like 30,000 lines of code. That would have taken me, I don't know, maybe a year to build," he said. "It wasn't overnight, but I probably spent 100 hours on that." Blomfield said he expects AI coding to radically change the software industry. "Instead of having coding assistance, we're going to have actual AI coders and then an AI project manager, an AI designer and, over time, an AI manager of all of this. And we're going to have swarms of these things," he said. Where people fit into this, he said, "is the question we're all grappling with." In 2021, Blomfield said in a podcast that would-be start-up founders should, first and foremost, learn to code. Today, he's not sure he'd give that advice because he thinks coders and software engineers could eventually be out of a job. "Coders feel like they are tending, kind of, organic gardens by hand," he said. "But we are producing these superhuman agents that are going to be as good as the best coders in the world, like very, very soon." The article includes an alternate opinion from Adam Resnick, a research manager at tech consultancy IDC. "The vast majority of developers are using AI tools in some way. And what we also see is that a reasonably high percentage of the code output from those tools needs further curation by people, by experienced people." NPR ends their article by noting that this further curation is "a job that AI can't do, he said. At least not yet."

Read more of this story at Slashdot.