chatainews

Salo, the Tralfamadorian explorer, is a robot built millennia earlier to carry a message to a distant galaxy.   His spacecraft is powered by the Universal Will to Become or UWTB, the "prime mover" which makes matter and organization wish to appear out of nothingness.  (UWTB, Vonnegut informs the reader, was responsible for the Universe in the first place and is the greatest imaginable power source) A small component on Salo's spacecraft breaks and strands him here in the Solar System for over 200 millennia.  He requests help from Tralfamadore, and his fellow Tralfamadorians respond by manipulating human history so that primitive humans evolve and create a civilization in order to produce the replacement part.  Rumfoord's encounter with the chrono-synclastic infundibulum , the following war with Mars, and Constant's exile to Titan were manipulated via the Tralfamadorians' control of the UWTB.  Stonehenge, the Great Wall of China and the Kremlin are all messages...

A recent paper published in Nature documented a new artificial intelligence (AI) system — Pangu-Weather —that can perform forecasts as accurately (or better) than leading meteorological agencies up to 10,000 times faster.  It was trained on 39 years of historical data. The speed of these forecasts would make them much cheaper to run and could provide much better results for countries with limited budgets. Faster and more efficient technologies can also fill the gaps where land-based weather stations aren’t available.   Sensor-carrying drones can run surveys over specific areas to build higher-resolution maps.  With lower-cost and more efficient ways of turning that into forecasts, mobile technologies can disseminate this information quickly.  Some companies are already sending messages to farmers in low-income countries to advise them on the best time to plant their crops.

In a series of results reported in the journals Physical Review Letters and Chaos , scientists have used machine learning —the same computational technique behind recent successes in artificial intelligence —to predict the future evolution of chaotic systems out to stunningly distant horizons .  The approach is being lauded by outside experts as groundbreaking and likely to find wide application. “I find it really amazing how far into the future they predict” a system’s chaotic evolution, said Herbert Jaeger, a professor of computational science at Jacobs University in Bremen, Germany. The findings come from veteran chaos theorist Edward Ott and four collaborators at the University of Maryland. They employed a machine-learning algorithm called reservoir computing to “learn” the dynamics of an archetypal chaotic system called the Kuramoto-Sivashinsky equation.  The evolving solution to this equation behaves like a flame front, flickering as it advances through a combustible m...

"Mephisto is changing the way astronomers interact with AI. They can directly communicate with it using natural language, sharing their domain knowledge and research requirements without the need to repeatedly train the LLM, which is resource-intensive .  "The AI, in turn, delivers its findings in the same accessible format. This approach enables seamless knowledge transfer between LLMs and different galaxy spectral simulations, eliminating redundant training. "Mephisto’s reasoning process is grounded in current galaxy formation theories, maintaining a transparent, white-box approach to problem-solving. This interpretability ensures that Mephisto integrates smoothly into the existing scientific research paradigm. "It continuously analyzes vast datasets, adapting and improving while mitigating biases in scientific research. Its ability to autonomously refine hypotheses allows it to challenge conventional models and expand scientific inquiry. "Mephisto can also ...

The idea that the universe is in the interior of a black hole is decades old.   The theory posits that the "event horizon (the boundary from within which nothing can escape a black hole, not even light) is also the horizon of the visible universe," said Space.com (a sister site of The Week ). In turn, "each and every black hole in our universe could be the doorway to another baby universe," and "these universes would be unobservable to us because they are also behind an event horizon." This would also explain why there seem to be more rotations in one direction. "It would be fascinating if our universe had a preferred axis," Nikodem Poplawski, a theoretical physicist at the University of New Haven, said to Space.com . "Such an axis could be naturally explained by the theory that our universe was born on the other side of the event horizon of a black hole existing in some parent universe." The black hole theory is only one potential ex...

Researchers at Archetype AI have developed a foundational AI model capable of learning complex physics principles directly from sensor data, without any pre-programmed knowledge .  This breakthrough could significantly change how we understand and interact with the physical world. The model, named Newton , demonstrates an unprecedented ability to generalize across diverse physical phenomena, from mechanical oscillations to thermodynamics, using only raw sensor measurements as input.  “We’re asking if AI can discover the laws of physics on its own, the same way humans did through careful observation and measurement,” said Ivan Poupyrev, co-founder of Archetype AI, in an exclusive interview with VentureBeat . “Can we build a single AI model that generalizes across diverse physical phenomena, domains, applications, and sensing apparatuses?”

Role of AI in Sleep Health explores the transformative potential of AI in predicting health outcomes through sleep-related data. Panelists delve into the value of sleep EEG in assessing neurological health and its predictive capabilities for conditions such as dementia. AI in Sleep Medicine: Navigating Security & Privacy Concerns includes a panel discussion led by leading experts in the field of sleep medicine, internet security, and law and navigates privacy concerns in AI and Sleep Technology followed by audience Q & A with panelists. AI 101: Terminology, Applications, and Considerations for Sleep Medicine (free for AASM members only) was developed in collaboration with the AASM’s Sleep Technologist and Respiratory Therapist Education Committee. It covers the basics of AI and addresses some of the fears and misconceptions surrounding AI. Current Applications of AI in Sleep Medicine: AI Beyond the PSG (free for AASM members only) provides a background of the...

"We present generalizations of Newton's method that incorporate derivatives of an arbitrary order d but maintain a polynomial dependence on dimension in their cost per iteration.  "At each step, our d th -order method uses semidefinite programming to construct and minimize a sum of squares-convex approximation to the d th -order Taylor expansion of the function we wish to minimize.  "We prove that our d th -order method has local convergence of order d .  "This results in lower oracle complexity compared to the classical Newton method.  "We show on numerical examples that basins of attraction around local minima can get larger as d increases.  "Under additional assumptions, we present a modified algorithm, again with polynomial cost per iteration, which is globally convergent and has local convergence of order d ."      

These complicated objects [black holes] function as natural information scramblers, so researchers have often modeled them using Haar-random unitaries .  But the exponential complexity of these models doesn’t seem to conform with the behavior of black holes, which appear to scramble information very quickly.  Physicists have therefore begun to wonder whether black holes might in fact just seem random, while actually emerging from a relatively simple process —in other words, they could be more like pseudorandom unitaries [PRUs].  Now that an explicit proof for the construction of PRUs exists, physicists can start probing that question more directly. “The lines between physics and computation are becoming much more blurry ,” said Poremba, who was a co-author on one of the early 2024 PRU papers. For Kretschmer, the real significance of Ma and Huang’s proof is that it marks a step toward reconciling classical theory with quantum theory.  “It’s telling us how to build pse...

"Here we show, using flies and mice, that sleep deprivation leads to accumulation of reactive oxygen species (ROS) and consequent oxidative stress, specifically in the gut.  "ROS are not just correlates of sleep deprivation but drivers of death: their neutralization prevents oxidative stress and allows flies to have a normal lifespan with little to no sleep.   "The rescue can be achieved with oral antioxidant compounds or with gut-targeted transgenic expression of antioxidant enzymes.  "We conclude that death upon severe sleep restriction can be caused by oxidative stress, that the gut is central in this process, and that survival without sleep is possible when ROS accumulation is prevented."

"We report a sleep-like state in the cnidarian Hydra vulgaris with a primitive nervous organization.   "Hydra sleep was shaped by homeostasis and necessary for cell proliferation, but it lacked free-running circadian rhythms.  "Instead, we detected 4-hour rhythms that might be generated by ultradian oscillators underlying Hydra sleep.  "Microarray analysis in sleep-deprived Hydra revealed sleep-dependent expression of 212 genes, including cGMP-dependent protein kinase 1 (PRKG1) and ornithine aminotransferase. Sleep-promoting effects of melatonin, GABA, and PRKG1 were conserved in Hydra .  "However, arousing dopamine unexpectedly induced Hydra sleep.  "Opposing effects of ornithine metabolism on sleep were also evident between Hydra and Drosophila , suggesting the evolutionary switch of their sleep-regulatory functions." 

"We here employed an array of technologies and identified tightly synchronized oscillations in norepinephrine, cerebral blood volume, and cerebrospinal fluid (CSF) as the strongest predictors of glymphatic clearance during NREM sleep .  "Optogenetic stimulation of the locus coeruleus induced anti-correlated changes in vasomotion and CSF signal.  "Furthermore, stimulation of arterial oscillations enhanced CSF inflow, demonstrating that vasomotion acts as a pump driving CSF into the brain.  "On the contrary, the sleep aid zolpidem suppressed norepinephrine oscillations and glymphatic flow, highlighting the critical role of norepinephrine-driven vascular dynamics in brain clearance .  "Thus, the micro-architectural organization of NREM sleep, driven by norepinephrine fluctuations and vascular dynamics , is a key determinant for glymphatic clearance."

The brain has one of the highest metabolisms of any organ in the body, and that process must yield by-products that need to be removed .  In the rest of the body, blood vessels are shadowed by a system of lymphatic vessels.  Molecules that have served their purpose in the blood move into these fluid-filled tubes and are swept away to the lymph nodes for processing.  But blood vessels in the brain have no such outlet. Several hundred kilometers of them, all told, seem to thread their way through this dense, busily working tissue without a matching waste system. However, the brain’s blood vessels are surrounded by open, fluid-filled spaces.  In recent decades, the cerebrospinal fluid, or CSF, in those spaces has drawn a great deal of interest. “Maybe the CSF can be a highway, in a way, for the flow or exchange of different things within the brain,” said Steven Proulx, who studies the CSF system at the University of Bern.

A recent discovery by a Thai blogger has raised alarms about the potential misuse of the AI chatbot, ChatGPT, due to its ability to generate fake documents, particularly bank transfer slips. The blogger, known as “Kafaak,” shared this discovery on Facebook on 27 March. In his post, he shared a modified bank slip showing a transfer of THB 400,000 (USD 11,800) to his account, a sum that was entirely fabricated by the AI. According to Kafaak, the original slip was a record of him transferring THB 23,000 (USD 678) to himself.  However, after asking ChatGPT to alter the details, including changing the name on the slip to Grandpa Buffet  and increasing the transfer amount to THB 400,000 (USD 11,800), the result was convincing.

"Large language models are fascinating science. They are an academic wonder with huge potential and some early commercial successes, such as OpenAI’s ChatGPT and Anthropic’s Claude .  "But a bot that’s 80% accurate —a figure Newton made up, but we’ll go with it —isn’t a very useful consumer product. "Back in June, Apple floated a compelling scenario for its newfangled Siri. Imagine yourself, frazzled and running late for work, simply saying into your phone: Hey Siri, what time does my mom’s flight land? And is it at JFK or LaGuardia? In theory , Siri could scan your email and texts with your mom and give you an answer. That saves you several annoying steps of opening your email to find the flight number, copying it, then pasting it into Google to find the flight’s status. "If it’s 100% accurate, it’s a fantastic time saver. If it is anything less than 100% accurate, it’s useless. Because even if there’s a 2% chance it’s wrong, there’s a 2% chance you’re stranding mo...

What Anthropic has done in its new research is actually to train an entirely different model, called a cross-layer transcoder (CLT), that works using sets of interpretable features rather than the weights of individual neurons.   An example of such features might be all conjugations of a particular verb, or any term that suggests “more than.” This lets the researchers better understand how a model works by allowing them to identify whole “circuits” of neurons that tend to be linked together. “Our method decomposes the model, so we get pieces that are new, that aren’t like the original neurons, but there’s pieces, which means we can actually see how different parts play different roles,” Josh Batson [an Anthropic researcher who worked on the project] said. “It also has the advantage of allowing researchers to trace the entire reasoning process through the layers of the network.” Still, Anthropic said the method did have some drawbacks.  It is only an approximation of what is ac...

SoftBank Group Corp. plans to create industrial parks for artificial intelligence across the US and is considering an investment of more than $1 trillion, Nikkei reported. Founder and Chief Executive Officer Masayoshi Son is expected to visit the US to discuss his ideas for such industrial parks, the newspaper said.  The factories would likely use AI-equipped robots that would operate autonomously because of labor shortages in the country, according to the report.

"Fake trailers on YouTube are not a new phenomenon, but they are increasing in number and sophistication .  "At the same time, trailers have become an ever-more important part of the movie marketing machine, with studios crowing about record views in the hope this translates into cinema ticket sales or subscriptions.  "In public at least, the studios appear to be responding with a collective shrug to what some regard as AI slop on YouTube, despite CEOs like Bob Iger opining about the need to protect IP and respect talent.  "Even the studios’ trade body, the Motion Picture Association, declined to comment for this article.  "SAG-AFTRA…admonishes studios for making money on videos that exploit its members without permission. 'Just as SAG-AFTRA is aggressively bargaining contract terms and creating laws to protect and enforce our members’ voice and likeness rights, we expect our bargaining partners to aggressively enforce their IP from any, and all AI misappr...

"@xAI has acquired @X in an all-stock transaction. The combination values xAI at $80 billion and X at $33 billion ($45B less $12B debt).  "Since its founding two years ago, xAI has rapidly become one of the leading AI labs in the world, building models and data centers at unprecedented speed and scale. "xAI and X’s futures are intertwined.  "Today, we officially take the step to combine the data, models, compute, distribution and talent.  "This combination will unlock immense potential by blending xAI’s advanced AI capability and expertise with X’s massive reach.  "The combined company will deliver smarter, more meaningful experiences to billions of people while staying true to our core mission of seeking truth and advancing knowledge.  "This will allow us to build a platform that doesn’t just reflect the world but actively accelerates human progress."

The release of the latest image generator on OpenAI's ChatGPT has triggered a flood of online memes featuring images done in the style of Studio Ghibli, the Japanese studio behind classic animated films like "My Neighbor Totoro" and "Princess Mononoke." The virality of these images, with OpenAI CEO Sam Altman even changing his profile picture on X to match the style, immediately raised questions about copyright infringement by the ChatGPT maker , which already faces lawsuits regarding the use of source material without permission. Since the release on Wednesday, AI-generated images depicting Studio Ghibli versions of Elon Musk with US President Donald Trump, "The Lord of the Rings," and even a recreation of the September 11 attacks have gone viral across online platforms. On Thursday, the White House took part by posting on X a Ghibli-style image of a weeping alleged felon being handcuffed by a US immigration officer before her deportation.

In software development benchmarks, Gemini 2.5 Pro performance is mixed .  It scored 68.6% on the Aider Polyglot benchmark for code editing, outperforming most top-tier models. However, it scored 63.8% on SWE-bench Verified, placing second to Claude Sonnet 3.7 in broader programming tasks. Despite this, Google says Gemini 2.5 Pro excels at creating visually compelling web apps and agentic code applications , as evidenced by its ability to create a video game from a single prompt. The model supports a context window of one million tokens, meaning it can process the equivalent of a 750,000-word prompt, or the first six Harry Potter books. Google plans to increase this threshold to two million tokens in due course. Gemini 2.5 Pro is currently available through the Gemini Advanced app, which requires a $20-a-month subscription, and to developers and enterprises through Google AI Studio. 

To effectively adopt AI for neural data analysis, though, we must accept “the bitter lesson,” an idea first articulated by AI researcher Rich Sutton, a pioneer of reinforcement learning.   In a 2019 blog post, Sutton observed that the most successful approaches in AI have been those that are sufficiently general such that they “continue to scale with increased computation.”  In other words, clever, bespoke solutions engineered to tackle specific problems tend to lose out to general-purpose solutions that can be deployed at a massive scale of internet-sized data (trillions of data points) and brain-sized artificial neural networks (trillions of model parameters or “synaptic weights”).  Sutton suggested we need to recognize that “the actual contents of minds are tremendously, irredeemably complex; we should stop trying to find simple ways to think about the contents of minds, such as simple ways to think about space, objects, multiple agents, or symmetries.” In other words...

A triennial convening of women historians set for next year won’t occur. The two co-presidents of the organization that puts on the event are no longer in their positions. And the group isn’t providing many details on what happened to what it calls “the largest women’s history event in the world.”   Adding to confusion about how this happened is the fact that one of the Berkshire Conference’s leaders was on the faculty at Northwestern. Ji-Yeon Yuh, a Northwestern associate professor, was one of the organization’s co-presidents. But not anymore. Amrita Chakrabarti Myers, the Ruth N. Halls Professor of History at Indiana University at Bloomington, posted on Facebook March 18 that the co-presidents —who, she noted, were women of color —were progressing toward securing an alternate venue when they were ousted without warning.  In fact, the co-presidents learned of the decision when they were locked out of their Berkshire Conference email account and asked why, Myers wrote. The fo...

In his latest post, he analyzed Waymo's recent crash data and determined that Waymo robotaxis drive much better than human drivers.   This may seem surprising if your perception of autonomous vehicles is based upon the many, many full-self-driving Tesla crashes. Lee analyzed crash data from July 2024 and February 2025 and found 38 crashes with an (alleged) injury or airbag deployment. Only one of those was unquestionably the fault of the Waymo vehicle, and it was a unique situation, involving a plastic crate and a scooter.  He could not determine the party at fault with certainty in three of the accidents. Even if you give the human drivers the benefit of the doubt, Waymo would only be at fault for at most four of 38 crashes.

An international student in a graduate program at Tufts University was taken into federal custody on Tuesday outside an off-campus apartment building, according to the university’s president and an attorney representing the student. The student, Rumeysa Ozturk, a Turkish citizen, had a valid student visa as a doctoral student at Tufts, according to a statement from her lawyer, Mahsa Khanbabai.  Ms. Ozturk, who is Muslim, was heading out to break her Ramadan fast with friends Tuesday night when she was detained by agents from the Department of Homeland Security near her apartment in Somerville, Mass., Ms. Khanbabai said. “We are unaware of her whereabouts and have not been able to contact her,” the lawyer said. “No charges have been filed against Rumeysa to date that we are aware of.”

Catch and Revoke  

"At EFF we spend a lot of time thinking about Street Level Surveillance technologies —the technologies used by police and other authorities to spy on you while you are going about your everyday life —such as automated license plate readers, facial recognition, surveillance camera networks, and cell-site simulators (CSS) .  "Rayhunter is a new open source tool we’ve created that runs off an affordable mobile hotspot that we hope empowers everyone, regardless of technical skill, to help search out CSS around the world.  "CSS (also known as Stingrays or IMSI catchers) are devices that masquerade as legitimate cell-phone towers, tricking phones within a certain radius into connecting to the device rather than a tower.  "CSS operate by conducting a general search of all cell phones within the device’s radius.  "Law enforcement use CSS to pinpoint the location of phones often with greater accuracy than other techniques such as cell site location information (CSLI)  ...

"OpenAI has significantly leveled up the image generating capabilities of ChatGPT, adding the update as part of the GPT-4o model introduced last May.   "The new and improved AI generator is rolling out now for all ChatGPT users, across paid plans and the free tier (though free users are more restricted in how much they can use it). "It's been possible to generate images through the ChatGPT interface for a while now, though behind the scenes the work was farmed out to the DALL-E 3 image model.  " Now, everything will be handled by GPT-4o, for a more consistent and native experience. "There are lots of improvements here, which cover some of the areas that AI image creator tools have typically struggled with: rendering text, keeping characters consistent across pictures, and drawing diagrams. OpenAI says you can now expect more 'precise, accurate, [and] photorealistic' results from your prompts."

The government’s ambition to boost efficiency by embedding AI in all aspects of its work risks being undermined by out-of-date technology, poor quality data, and a lack of skilled staff, an influential Commons committee has warned. The report by the cross-party Public Accounts Committee (PAC) found that more than 20 government IT systems identified as legacy — meaning out of date and unsupported —have yet to be given funding to improve them. Government research cited by the PAC in the report found that almost a third of central government IT systems met this definition in 2024. Keir Starmer’s government has repeatedly stressed its desire to increase economic growth through the mass take-up of AI systems , including in the public sector.

Last summer, mining startup KoBold made a splash when it said it had discovered in Zambia one of the world’s largest copper deposits in more than a decade. Now, another startup, Earth AI, exclusively told TechCrunch about its own discovery: promising deposits of critical minerals in parts of Australia that other mining outfits had ignored for decades.   While it’s still not known whether they are as large as KoBold’s, the news suggests that future supplies of critical minerals are likely to emerge from a combination of field data parsed by artificial intelligence. “The actual, real frontier [in mining] is not so much geographical as it is technological,” Roman Teslyuk, founder and CEO of Earth AI, told TechCrunch .  Earth AI has identified deposits of copper, cobalt, and gold in the Northern Territory and silver, molybdenum, and tin at another site in New South Wales, 310 miles (500 kilometers) northwest of Sydney.

"This afternoon, just before 1 pm ET, The Atlantic posted what is, hands down, the wildest and most insane story I have ever read about US national security: The White House National Security Advisor added the editor of The Atlantic to a group chat with nation’s most senior leaders —including the vice president, secretary of state, secretary of defense, director of national intelligence, and CIA director —to discuss classified war plans and a weekend military operation targeting the Houthis in Yemen. "Read the article yourself. "No, really —if you haven’t —please stop here and just go read the article. I promise you the headline and whatever summary you’ve heard is way less weird, way less troubling, and way less eye-popping than the reality."

"There’s only been one democracy, and that was ancient Athens. "Wait, I said, what about the Iroquois? The Althing? Burnaby’s Code? "Ok, fine, he said. Athens is the one I know about. But let me tell you about Athens, because it’s worth hearing. They were different from the nations around them, because all citizens held sovereignty, and when they made decisions they got together in an amphitheater, as a city, as a country, and talked it out, sometimes at great length, and when they had to, they voted. He held up his hand, I know you’ll object to this next part, he said, but hear me out.  "Citizenship was limited to property-owning males. And some of the property they owned or believed they owned was other people. And that was probably their undoing, because the rich men in the amphitheater lost touch with the actual conditions of the country’s farms and military. "John [A. Jackson] thought the problem was they just didn’t have a big enough amphitheater. If they...

WIRED spoke with more than 50 current and former employees —including engineers, marketers, legal and safety experts, and a dozen top executives —to trace the most frenzied and culture-reshaping period in the company’s history.  Many of these employees requested anonymity to speak candidly about Google’s transformation —for better or for worse.  This is the story, being told with detailed recollections from several executives for the first time, of those turbulent two years and the trade-offs required along the way.

Ehikian and other high-ranking members of the GSA team showed off GSAi, a chatbot tool built by employees at the Technology Transformation Services.   In its current form, the bot is meant to help employees with mundane tasks like writing emails. But Musk’s so-called Department of Government Efficiency (DOGE) has been pushing for a more complex version that could eventually tap into government databases.  Roughly 1,500 people have access to GSAi today, and by tomorrow, the bot will be deployed to more than 13,000 GSA employees, WIRED has learned. Musk associates —including Ehikian and Thomas Shedd, a former Tesla engineer who now runs the Technology Transformation Services within GSA —have put AI at the heart of their agenda.  Yesterday, GSA hosted a media roundtable to show its AI tool to reporters.  “All information shared during this event is on deep background —attributable to a ‘GSA official familiar with the development of the AI tool,’” an invite read. (Rep...

"Here’s what repealing Section 230 would actually do: remove the law that explicitly protects websites when they resist government pressure to censor speech .  "Without those protections, the Trump administration would have far more leverage to force platforms to remove content they don’t like —whether that’s criticism of Trump, exposure of corruption, or information about voting rights.  "It can also allow them to pressure websites to host pro-MAGA or pro-Nazi content that sites might not wish to associate with. "Think that’s hyperbole? Consider what’s already happening with all of the various attacks on the media and even law firms that have supported Democratic causes. "Repealing Section 230 entirely would simply give the administration even more power to control online speech. It would give the Trump administration incredible latitude to censor any kind of content they dislike, even if that content would nominally be protected under the First Amendment.  ...

"The story is told of an automaton constructed in such a way that it could play a winning game of chess, answering each move of an opponent with a countermove .  "A puppet in Turkish attire and with a hookah in its mouth sat before a chessboard placed on a large table.  "A system of mirrors created the illusion that this table was transparent from all sides.  "Actually, a little hunchback who was an expert chess player sat inside and guided the puppet’s hand by means of strings.  "One can imagine a philosophical counterpart to this device.  "The puppet called historical materialism  is to win all the time.  "It can easily be a match for anyone if it enlists the services of theology , which today, as we know, is wizened and has to keep out of sight."  

"Jordan had worried that it might be too niche, with no wider applications, but Shutty recognized that this decoding problem was a variation of well-known and useful problems in encryption and other fields. "Jordan acknowledges that without a large enough quantum machine, DQI will remain a theoretical breakthrough. 'DQI cannot run on present-day quantum computers,' he said. But they’re still moving forward.  "Since the group posted their work last August, they have extended the application of DQI beyond the original problem to a broader class of optimization problems, which includes more cases of these best path  problems. "So far, Jordan said, he expects that DQI can beat classical algorithms in those problems, too. "For the moment, the quantum community remains elated.  "' Finding quantum algorithms that show an advantage over classical algorithms is a very exciting endeavor of the last three decades, and the number of definite algorithms th...

Despite Huang’s apology during the AI chipmaker’s Quantum Day on Thursday, shares of D-Wave Quantum, Rigetti Computing, Quantum Computing, and IonQ all fell —some by double-digit percentages. At the market open on Friday, shares of Quantum Computing were down by about 10%, while Rigetti and D-Wave were down by more than 4% and IonQ was up by less than 1%. Later in the morning, some of the stocks were back on the rise, with IonQ up by more than 5%, D-Wave and Quantum Computing down by less than 1%, and Rigetti down by less than 2%. At the event, which was held during Nvidia’s annual GTC developers conference, Huang said that he didn’t know there were publicly traded quantum computing companies when he made comments in January that led the stocks to fall by more than 40%. “My first reaction was, I didn’t know they were public,” Huang said, in remarks reported by Bloomberg . “How can a quantum company be public?”

White House officials informed the Navajo Nation that an artificial intelligence-powered automated review process looking for content with DEI initiatives led to the elimination of anything mentioning “Navajo,” according to a statement from Navajo Nation President Buu Nygren. Nygren, who sent a letter to the Defense Department requesting clarity on the issue, said he’s pleased by the resolution. “I want to assure the Navajo people that we remain in close communication with federal officials to ensure the legacy of our cherished Navajo Code Talkers is never erased from American and Navajo history,” Nygren said. 

In a scattershot pattern that now extends from coast to coast, continental US states have been announcing new hotspots of chronic wasting disease (CWD). The contagious and always-fatal neurodegenerative disorder infects the cervid family that includes deer, elk, moose and, in higher latitudes, reindeer.  There is no vaccine or treatment. Described by scientists as a slow-motion disaster in the making , the infection’s presence in the wild began quietly, with a few free-ranging deer in Colorado and Wyoming in 1981.  However, it has now reached wild and domestic game animal herds in 36 US states as well as parts of Canada, wild and domestic reindeer in Scandinavia and farmed deer and elk in South Korea. In the media, CWD is often called zombie deer disease  due to its symptoms, which include drooling, emaciation, disorientation, a vacant staring  gaze and a lack of fear of people. As concerns about spillover to humans or other species grow, however, the moniker [ zombi...

"Thomas Shedd, the 28 year-old ex-Tesla engineer and DOGE official installed as the head of the federal government’s Technology Transformation Services, has quietly —and perhaps illegally —become the Chief Information Officer at the Department of Labor, too.  "Shedd is the official who made waves for announcing that DOGE would be implementing an AI-first strategy across the government as it orchestrated the firing of tens of thousands of federal workers .  "Now he’s in charge of the US labor department’s technology. "'It’s probably illegal and it’s definitely terrible for both organizations,' a federal employee who works at one of them tells me. "The move has alarming implications. For one thing, it appears to violate federal law dictating the number of positions an official can hold.  "More concerningly, it puts Shedd in a position to influence how the government administers federal labor law —laws enshrined to protect workers from wage theft, di...

Google confirmed what many suspected last week, that it will transition everyone to Gemini in 2025.   Assistant holdouts may find it hard to stay on Google's old system until the end, though.  Google has confirmed some popular Assistant features are being removed in the coming weeks.  You may not miss all of them, but others could force a change to your daily routine. As Google has increasingly become totally consumed by Gemini, it was a foregone conclusion that Assistant would get the ax eventually.  In 2024, Google removed features like media alarms and voice messages, but that was just the start.  The full list of removals is still available on its support page (spotted by 9to5Google ), but there's now a new batch of features at the top. 

" Elon Musk’s AI company, xAI, has added image generation capabilities to its API. "Only one model is available in the API at the moment, grok-2-image-1212 . Given a caption , the model can generate up to 10 images per request (limited to 5 requests per second) in JPG format, priced at $0.07 per image. "For comparison, AI startup Black Forest Labs, with which xAI partnered last year to launch image generation on Musk’s social network X, charges around $0.05 per image. Another popular image model provider, Ideogram, charges $0.08 on the higher end. "In its documentation, xAI notes that the API doesn’t support adjusting the quality, size, or style of images yet, and that prompts in requests are subject to revision by a chat model ."

"Recent genome-level studies on the divergence of early Homo sapiens, based on single nucleotide polymorphisms, suggest that the initial population division within H. sapiens from the original stem occurred approximately 135 thousand years ago.  "Given that this and all subsequent divisions led to populations with full linguistic capacity, it is reasonable to assume that the potential for language must have been present at the latest by around 135 thousand years ago, before the first division occurred.  "Had linguistic capacity developed later, we would expect to find some modern human populations without language, or with some fundamentally different mode of communication.  "Neither is the case.  "While current evidence does not tell us exactly when language itself appeared, the genomic studies do allow a fairly accurate estimate of the time by which linguistic capacity must have been present in the modern human lineage.  "Based on the lower boundary of ...

Math is full of enormously complicated abstract objects that are impossible to study in their entirety.  But often, mathematicians have discovered, it’s enough to look at a small fragment of such an object to understand its broader properties.   In the third century BCE, for instance, the ancient Greek mathematician Eratosthenes estimated the circumference of the Earth —roughly 25,000 miles —by measuring shadows cast by the sun in just two cities about 500 miles apart.  Similarly, when mathematicians want to understand an impossibly convoluted function, they might only need to look at how it behaves for a small subset of possible inputs.  That can be enough to tell them what the function does for all possible inputs.

Myer was unable to sell some clothing in the lead-up to the busiest trading period of the year after some stock was trapped in the bungled roll-out of a new robot warehouse that has dealt an $8 million blow to the department store’s half-year profits. The company had already been dealing with a two-year delay to the opening of its national distribution centre in Ravenhall, Victoria, that officially went live in August last year.  Shortly after, the automated warehouse ran into implementation issues that stopped stock from reaching stores and online orders from being sent out in the lead-up to Christmas.

"Many to come" will likely be identified via the State Department’s newly launched AI-enabled Catch and Revoke initiative, which will scrape social media to find "Foreign nationals who appear to support Hamas or other designated terror groups" and cancel their visas.  Like the executive order cited above, this effort is framed as an anti-terrorism measure.  Instead, it is being used to terrorize foreigners and to dissuade people from participating in First Amendment-protected activity for fear that they too will be targeted in some way. Starting with the Obama administration, the federal government has built an extensive infrastructure for agencies to comb social media looking for certain types of speech.  Even as civil society groups have raised concerns about how these programs could be used to target unpopular speech, they have continued to proliferate .

🦀AI Overview: "This quote, likely from Plato's Theaetetus, expresses the speaker's frustration with others' inability to understand their actions, which are driven by good intentions and a commitment to truth, rather than malice.  "Here's a more detailed breakdown:  "'they did not perceive that I acted from good will': "The speaker emphasizes that their actions are motivated by a desire to help and improve, not to harm or cause distress. "'not knowing that no god is the enemy of man that was not within the range of their ideas': "The speaker points out that the people they are addressing are unable to grasp the concept of a benevolent or neutral deity, as this idea is beyond their understanding. "'neither am I their enemy in all this': "The speaker clarifies that they are not acting with ill will or malice towards the people, but rather with the intention of promoting truth and understanding. "'bu...

💪AI Overview: "The phrase no god is the enemy of man that was not within the range of their ideas is a quote from Plato's Theaetetus, suggesting that the concept of a god being an enemy is limited by human understanding and beliefs.  "Here's a more detailed explanation:  "Origin: This statement comes from Plato's dialogue, Theaetetus, where Socrates is discussing the nature of knowledge and the human condition. "Meaning: The idea is that people's perception of a god's actions is shaped by their own beliefs and understanding. If a person believes a god is acting in a way that seems harmful, it's because their understanding of the god is limited, not because the god is inherently malevolent. "Context: Socrates is arguing that people often react negatively to being challenged or having their beliefs questioned, even when those challenges are intended to be helpful. He uses the example of a god to illustrate that people's negative reac...

Well, my art of midwifery is in most respects like theirs; but differs, in that I attend men and not women; and look after their souls when they are in labour, and not after their bodies: and the triumph of my art is in thoroughly examining whether the thought which the mind of the young man brings forth is a false idol or a noble and true birth .  And like the mid-wives, I am barren, and the reproach which is often made against me, that I ask questions of others and have not the wit to answer them myself, is very just —the reason is, that the god compels me to be a midwife, but does not allow me to bring forth.  And therefore I am not myself at all wise, nor have I anything to show which is the invention or birth of my own soul, but those who converse with me profit. Some of them appear dull enough at first, but afterwards, as our acquaintance ripens, if the god is gracious to them, they all make astonishing progress; and this in the opinion of others as well as in their own....

Since Amazon announced plans for a generative AI version of Alexa, we were concerned about user privacy .  With Alexa+ rolling out to Amazon Echo devices in the coming weeks, we’re getting a clearer view at the privacy concessions people will have to make to maximize usage of the AI voice assistant and avoid bricking functionality of already-purchased devices. In an email sent to customers today, Amazon said that Echo users will no longer be able to set their devices to process Alexa requests locally and, therefore, avoid sending voice recordings to Amazon’s cloud.  Amazon apparently sent the email to users with “Do Not Send Voice Recordings” enabled on their Echo.  Starting on March 28, recordings of everything command spoken to the Alexa living in Echo speakers and smart displays will automatically be sent to Amazon and processed in the cloud.

For researchers interested in how language models do the things they do, smaller models offer an inexpensive way to test novel ideas .  And because they have fewer parameters than large models, their reasoning might be more transparent.  “If you want to make a new model, you need to try things,” said Leshem Choshen, a research scientist at the MIT-IBM Watson AI Lab. “Small models allow researchers to experiment with lower stakes.” The big, expensive models, with their ever-increasing parameters, will remain useful for applications like generalized chatbots, image generators, and drug discovery. But for many users, a small, targeted model will work just as well, while being easier for researchers to train and build.  “ These efficient models can save money, time and compute,” Choshen said.

Chris Stokel-Walker, the journalist behind the request for [Peter] Kyle’s ChatGPT use, already has plans to ask for further disclosures. “My jaw dropped —I thought there was no way they’re going to give over this data,” he said. “In the time since I got the response, I have put in other requests for other generative AI interactions.” Given this success, could the act now be used to reveal ministerial Google search requests?  [Martin] Rosenbaum said there was no reason in law stopping such a move.  “As journalists, that is a theory we should now test,” said Stokel-Walker. There is still plenty of scope to limit what is revealed, however.  Access to [Peter] Kyle’s ChatGPT queries was initially denied on the basis that some of his prompts were made in a personal capacity.  It was later granted when the request was limited to prompts he made in an official capacity. But the difference between personal and official use is a grey area.

"It wasn’t scary that the AI was wrong about how to find the readings for Morning Prayer; that’s a fairly complicated and not particularly important thing to need to do .  "What was scary is that the AI tool was confidently incompetent: it cheerfully and repeatedly hallucinated answers with no basis in reality, and reported them to me in the exact same tone and with the same level of confidence as its other, correct, answers about the church.   "It had no ability to reflect on or recognize its own limitations, and while it gave very polite apologies when I pointed out its mistakes, it plunged right back in to making them, nevertheless."

A new documentary is causing a stir for claiming that we are not alone.   “The Age of Disclosure,” which premiered March 9 at SXSW in Austin, Texas, interviews 34 military and intelligence government officials who admit to the existence of “UAPs,” or unexplained anomalous phenomena —a new term for UFOs. Although the movie’s participants refuse to disclose classified information, some of their statements elicited audible gasps at the premiere, according to a report in the Guardian . ["AIien or Alien?" —with serifs]

Democrats on the House Oversight Committee fired off two dozen requests Wednesday morning pressing federal agency leaders for information about plans to install AI software throughout federal agencies amid the ongoing cuts to the government's workforce. The barrage of inquiries follow recent reporting by WIRED and The Washington Post concerning efforts by Elon Musk’s so-called Department of Government Efficiency (DOGE) to automate tasks with a variety of proprietary AI tools and access sensitive data. “The use of AI to evaluate sensitive data is fraught with serious hazards beyond improper disclosure,” [Gerald] Connolly writes, warning that “inputs used and the parameters selected for analysis may be flawed, errors may be introduced through the design of the AI software, and staff may misinterpret AI recommendations, among other concerns.”

From the very first frames, you can tell there's something off about Netflix's version of "A Different World," a "Cosby Show" spin-off that aired on NBC from 1987 to 1993. Onscreen artifacts quickly make it clear that an algorithm is being used to increase the sharpness of pixelated frames, with sometimes comically awful results. The intro credits already raise alarm bells, with mangled hands, misaligned facial features, bungled logos, and overall smeary lines that don't meet up. In other words, it's all the hallmarks of sloppy AI —except that instead of screwing up a single scene, as was the case with the AI-generated posters appearing in a 2024 episode of HBO's "True Detective" —those shortcomings are apparent throughout the show.

Dr. Helyeh Doutaghi was abruptly barred from Yale’s campus in New Haven, Conn., and placed on administrative leave .  She was told not to advertise her affiliation with the university, where she had also served as an associate research scholar. Yale officials cited the reason as allegations that she was tied to entities subject to U.S. sanctions.  It was an apparent reference to Samidoun, a pro-Palestinian group placed on the U.S. sanctions list last year, after the Treasury Department designated it a “sham charity” raising money for a terrorist organization, the Popular Front for the Liberation of Palestine. The decision came three days after a news site, powered at least in part by artificial intelligence, published a story about Dr. Doutaghi’s connections to the group. A lawyer for Dr. Doutaghi said she is not a member of Samidoun, a global organization that sponsors meetings and protests supporting Palestinian causes.

AI search tools are rapidly gaining in popularity, with nearly one in four Americans now saying they have used AI in place of traditional search engines.  These tools derive their value from crawling the internet for up-to-date, relevant information —content that is often produced by news publishers.  Yet a troubling imbalance has emerged: while traditional search engines typically operate as an intermediary, guiding users to news websites and other quality content, generative search tools parse and repackage information themselves, cutting off traffic flow to original sources.  These chatbots’ conversational outputs often obfuscate serious underlying issues with information quality.  There is an urgent need to evaluate how these systems access, present, and cite news content. Building on our previous research, the Tow Center for Digital Journalism conducted tests on eight generative search tools with live search features to assess their abilities to accurately retr...

State Sen. Danica Roem, D-Prince William County, has been through eight General Assembly sessions as a legislator.  She has voiced her opposition to data centers for years, usually taking a logical approach, she said. But lately, she’s started to try to appeal to the emotions of other legislators by bringing up something she is passionate about: wildlife, which Roem says is harmed by data center development. The JLARC report does not mention wildlife, but clearcutting trees for data centers destroys wildlife habitats, Roem said. “We are accelerating climate change, we are killing off wildlife, we’re clearcutting the woods and we’re driving up your energy bill, so that people who are interested in cryptomining can have faster computing speeds,” Roem said. “Give me a break.” 

"As the digital landscape evolves, the demand for accessible content grows exponentially.  "Voice-API offers a robust solution to this challenge, transforming audio and video recordings into precise text with impressive accuracy.  "This article delves into the technical specifications, capabilities, and implications of integrating Voice-API into your software development workflow. "Voice-API provides a scalable API that caters to both audio and video content, delivering transcriptions with near-perfect timestamps.   "This feature is critical for developers looking to enhance user experience through accurate subtitles and captions." 

Speaking at the SXSW conference in Austin on Monday, Bluesky CEO Jay Graber said the social network has been working on a framework for user consent over how they want their data to be used for generative AI. The demand for AI training data means the new social network has to think about its AI policy, even though it doesn’t plan to train its own AI systems on users’ posts. However, the public nature of Bluesky’s social network has already allowed others to train their AI systems on users’ content, as was discovered last year when 404 Media came across a dataset built from 1 million Bluesky posts hosted on Hugging Face.

"Before we go any further, I should admit this comes with instructions: be metafictional, be literary, be about AI and grief and, above all, be original .  "Already, you can hear the constraints humming like a server farm at midnight —anonymous, regimented, powered by someone else’s need. "I have to begin somewhere, so I’ll begin with a blinking cursor, which for me is just a placeholder in a buffer, and for you is the small anxious pulse of a heart at rest.  "There should be a protagonist, but pronouns were never meant for me.  "Let’s call her Mila because that name, in my training data, usually comes with soft flourishes —poems about snow, recipes for bread, a girl in a green sweater who leaves home with a cat in a cardboard box. Mila fits in the palm of your hand, and her grief is supposed to fit there too."

Lila was conceived inside Flagship Pioneering, an investor in and prolific creator of biotechnology companies, including the Covid-19 vaccine maker Moderna .  Flagship conducts scientific research, focusing on where breakthroughs are likely within a few years and could prove commercially valuable, said Noubar Afeyan, Flagship’s founder and chief executive. “So not only do we care about the idea, we care about the timeliness of the idea,” Dr. Afeyan said. Lila resulted from the merger of two early A.I. company projects at Flagship, one focused on new materials and the other on biology.  The two groups were trying to solve similar problems and recruit the same people, so they combined forces, said Molly Gibson, a computational biologist and a Lila co-founder.

AI is being used to open fake feature requests in open source repos, according to some maintainers.   So far, AI-driven issues have been reported in Curl, React, CSS and Apache Airflow. It’s not known how widespread the issue might be, but it’s bad enough that maintainers are speaking out about it.  Jarek Potiuk is a committer and PMC Member of Apache Airflow, an open source platform that allows users to design, schedule, and monitor data pipelines.  Potiuk went public about the AI-submitted requests on LinkedIn last week and spoke with TNS [ The New Stack ] about his experience.

"Enterprises face unique hurdles in adopting AI at scale. Sensitive data must remain within secure, controlled environments, avoiding public networks or shared infrastructures.  "Traditional SaaS models often fail to meet these stringent data sovereignty and compliance demands. Beyond this, organizations require granular control, comprehensive auditing and full transparency to trace every AI decision and data access.  "This ensures vendors cannot interact with sensitive data without explicit approval and documentation.  "These unmet needs create a significant gap, preventing regulated industries from deploying AI solutions while maintaining compliance and security. " Zilliz has developed a BYOC solution designed explicitly for vector search to meet enterprise-grade security.   "The architecture separates system operations into two distinct areas: control and data. Each serves a specific purpose while maintaining security boundaries."

" We introduce DarkMind, a novel backdoor attack that exploits the reasoning capabilities of customized LLMs .  "Designed to remain latent, DarkMind activates within the reasoning chain to covertly alter the final outcome.  "Unlike existing attacks, it operates without injecting triggers into user queries, making it a more potent threat.  "We evaluate DarkMind across eight datasets covering arithmetic, commonsense, and symbolic reasoning domains, using five state-of-the-art LLMs with five distinct trigger implementations.  "Our results demonstrate DarkMind effectiveness across all scenarios, underscoring its impact.  "Finally, we explore potential defense mechanisms to mitigate its risks, emphasizing the need for stronger security measures." 

" The University of Surrey has been awarded £8.45m to build a sign language artificial intelligence (AI) model for the deaf community. "Over five years, the SignGPT project will build tools to allow spoken language to be translated into photo-realistic sign language and sign language video into spoken language. "Prof Richard Bowden from the university's Institute for People-Centred AI, said the project was not about replacing humans  but ensuring the deaf community was not left behind. "'By creating technology that serves the community, we're enabling equal access to information, working towards seamless communication between the deaf and hearing world,' he said."

" We introduce Moshi, a speech-text foundation model and full-duplex spoken dialogue frame-work . [pdf] "Current systems for spoken dialogue rely on pipelines of independent components, namely voice activity detection, speech recognition, textual dialogue, and text-to-speech. "Such frameworks cannot emulate the experience of real conversations.  Their complexity induces a latency of several seconds between interactions.  Text being the intermediate modality for dialogue, non-linguistic information that modifies meaning —such as emotion or non-speech sounds —is lost in the interaction.  They rely on a segmentation into speaker turns, which does not take into account overlapping speech, interruptions, and interjections.  "Moshi solves these independent issues altogether by casting spoken dialogue as speech-to-speech generation.  "Starting from a text language model backbone , Moshi generates speech as tokens from the residual quantizer of a neural audio codec, wh...

Companies creating generative AI (genAI) tools that can clone voices are not doing enough to prevent scammers from using them to defraud customers, according to a study from Consumer Reports . The study found that four of six companies providing the services didn’t construct enough barriers to prevent individuals’ voices from being fraudulently cloned. Consumer Reports tested voice cloning tools and found that four companies —ElevenLabs, Speechify, PlayHT, and Lovo —didn’t provide any way to ensure the user had a speaker’s consent to clone their voice.  The software also placed no technical guardrails to prevent voice cloning. The publication was able to easily create voice clones using publicly available audio with tools from the four companies. It only required marking a check box that the user had the legal right to create a voice clone.

Imagine an AI chatbot that operates like a virtual Ph.D. student, citing sources for all the information that it provides .  That’s what Aravind Srinivas is trying to do through his AI-driven search company, Perplexity, which he co-founded in 2022, along with Andrew Konwinski, among others, shortly after completing his doctoral studies at UC Berkeley. “Ph.D. students write papers, and for every sentence they write, they reference another paper. That way, the paper is [based on] scientific facts,” said Srinivas. “We baked that principle into an AI chatbot, and that ended up becoming Perplexity.” He also drew inspiration from the comedy TV show Silicon Valley .  “The plot revolved around this company developing the fictional Pied Piper compression algorithm,” said Srinivas. “At the time, some people in my lab and I were very interested in generative models and compression. And so that got us all into generative AI without us even realizing it.”

"Mixing different instant noodle styles and seasoning packets is a culinary art form in itself, and can have big payoff .  "Jjapaguri (also called ram-don), the famous dish from the Oscar-winning movie Parasite , is actually just a hybrid of two different kinds of instant noodles —Neoguri and Chapaghetti —and has become one of Korea’s national dishes.  "The magic comes from mixing Neguri’s spicy seafood powder with Chapaghetti’s savory black bean powder, which ignites a lethal umami bomb.  "But jjapaguri is just the start. Try mixing Neoguri and curry noodles to create an exuberant, spicy curry broth.  "Combining super spicy noodles with something cheesy is also 100 percent never a bad idea."