chatainews

"Jeremy Nguyen, a senior researcher at Swinburne University of Technology, in Australia, ran an experiment last year to see how exposure to AI-generated text might change the way people write.  "He and his colleagues asked 320 people to write a post advertising a sofa for sale on a secondhand marketplace.  "Afterward, the researchers showed the participants what ChatGPT had written when given the same prompt, and they asked the subjects to do the same task again. The responses changed dramatically. "'We didn’t say, Hey, try to make it better, or more like GPT ,' Nguyen told me.  "Yet more like GPT  is essentially what happened: After the participants saw the AI-generated text, they became more verbose, drafting 87 words on average versus 32.7 in the first round. The full results of the experiment are yet to be published or peer-reviewed, but it’s an intriguing finding.  "Text generators tend to write long, even when the prompt is curt. Might peopl...

As of 2025, the Catholic Church has named 37 Doctors of the Church. Of these, the 18 who died before the Great Schism of 1054 are also held in high esteem by the Eastern Orthodox Church, although it does not use the formal title Doctor of the Church. Among the 37 recognised Doctors, 28 are from the West and nine from the East; four are women and thirty-three are men; one is an abbess, three are nuns, and one is a tertiary associated with a religious order; two are popes, 19 are bishops, twelve are priests, and one is a deacon; and 27 are from Europe, three are from Africa, and seven are from Asia.  More Doctors (twelve) lived in the fourth century than any other; eminent Christian writers of the first, second, and third centuries are usually referred to as the Ante-Nicene Fathers .  The shortest period between death and nomination was that of Alphonsus Liguori, who died in 1787 and was named a Doctor in 1871 —a period of 84 years; the longest was that of Irenaeus, which took m...

"A team from Beijing Normal University in China turned to a group of people who already had thin electrodes inserted into their brains as part of an experimental headache therapy, bypassing the ethical question of whether this sort of research justifies an invasive operation. "The researchers administered a visual perception test to these patients.  "A blinking object was displayed on a screen, which would hide itself for half of the time of the test.  'These characteristics meant the patients had to pay attention to the object and adjust their eyes and focus to keep watching it, rather than just regard the screen without analyzing it.  "This thus facilitated conscious perception, with the already-implanted electrodes then recording the brain activity that accompanied this. "The authors write: The findings indicate that the intralaminar and medial thalamic nuclei regulate conscious perception. This conclusion represents a significant advance in our underst...

"In a new joint study, researchers with OpenAI and the MIT Media Lab found that this small subset of ChatGPT users engaged in more problematic use , defined in the paper as indicators of addiction... including  Preoccupation,  Withdrawal symptoms,  Loss of control, and  Mood modification. "To get there, the MIT and OpenAI team surveyed thousands of ChatGPT users to glean not only how they felt about the chatbot, but also to study what kinds of affective cues , which was defined in a joint summary of the research as aspects of interactions that indicate empathy, affection, or support  they used when chatting with it. "Though the vast majority of people surveyed didn't engage emotionally with ChatGPT, those who used the chatbot for longer periods of time seemed to start considering it to be a friend .  "The survey participants who chatted with ChatGPT the longest tended to be lonelier and get more stressed out over subtle changes in the model's behavior, too....

Instead of depressing wages or taking jobs, generative AI chatbots like ChatGPT, Claude, and Gemini have had almost no wage or labor impact so far —a finding that calls into question the huge capital expenditures required to create and run AI models. In a working paper released earlier this month, economists Anders Humlum and Emilie Vestergaard looked at the labor market impact of AI chatbots on 11 occupations, covering 25,000 workers and 7,000 workplaces in Denmark in 2023 and 2024. Many of these occupations have been described as being vulnerable to AI: accountants, customer support specialists, financial advisors, HR professionals, IT support specialists, journalists, legal professionals, marketing professionals, office clerks, software developers, and teachers.

"Carnegie Mellon researchers instructed artificial intelligence models from Google, OpenAI, Anthropic, and Meta to complete tasks a real employee might carry out in fields such as finance, administration, and software engineering.  "In one, the AI had to navigate through several files to analyze a coffee shop chain's databases. In another, it was asked to collect feedback on a 36-year-old engineer and write a performance review. Some tasks challenged the models' visual capabilities: One required the models to watch video tours of prospective new office spaces and pick the one with the best health facilities. "The results weren't great: The top-performing model, Anthropic's Claude 3.5 Sonnet, finished a little less than one-quarter of all tasks. The rest, including Google's Gemini 2.0 Flash and the one that powers ChatGPT, completed about 10% of the assignments.  "The findings, along with other emerging research about AI agents, complicate the ide...

"Chinese e-commerce and web giant Alibaba’s Qwen team has officially launched a new series of open source AI large language multimodal models known as Qwen3 that appear to be among the state-of-the-art for open models, and approach performance of proprietary models from the likes of OpenAI and Google. "The Qwen3 series features two mixture-of-experts  models and six dense models for a total of eight (!) new models. "The mixture-of-experts  approach involves having several different specialty model types combined into one, with only those relevant models to the task at hand being activated when needed in the internal settings of the model (known as parameters). It was popularized by open source French AI startup Mistral. "According to the team, the 235-billion parameter version of Qwen3 codenamed A22B outperforms DeepSeek’s open source R1 and OpenAI’s proprietary o1 on key third-party benchmarks including ArenaHard (with 500 user questions in software engineering and...

OpenAI is updating ChatGPT search, its web search tool in ChatGPT, to give users an improved online shopping experience, the company announced Monday. When ChatGPT users search for products, the chatbot will now offer a few recommendations, present images and reviews for those items, and include direct links to web pages where users can buy the products. OpenAI says users can ask super-specific questions in natural language and receive customized results.  To start, OpenAI is experimenting with categories such as fashion, beauty, home goods, and electronics.

"A quantum computer represents information as a probability of one and zero, known as a superposition. Superposition is a concept from quantum mechanics: for example, an electron can exist as a superposition, or probability, of multiple locations. "You can also think of superposition like a coin flipping in the air. Before it lands, it is neither heads nor tails, but in a superposition state of both.  "Similarly, the qubit can represent information as some probability of both one and zero. "Researchers make physical qubits from different materials —for Google, Amazon, and IBM, each qubit is a small superconducting circuit; notable startups are using ions, atoms, and photons as qubits.  "At this point, it’s not clear what material is best. "All qubits obey the mathematics of quantum mechanics. So do molecules. That’s why experts predict that an early useful application of quantum computers could be performing accurate and fast chemistry simulations, for dis...

The Biden Administration moved aggressively in its final 18 months to convince more than 200 AI technology experts to forgo the private sector for the federal workforce, through what was called the National AI Talent Surge .  The new hires were deployed throughout the government and used AI to find ways to reduce Social Security wait times, simplify tax filings, and help veterans track their medical care.  Most of them were quickly pushed out by the new administration, multiple former federal officials tell TIME . The shift, say the former officials, represents an enormous waste of federal resources, as agencies across the Trump Administration are looking to draw workers with the very experience they just let go.  "It also means agencies may have to increasingly rely on costlier outside companies for that expertise."

" Katherine Bindley spoke to David Markley, an Amazon veteran turned executive coach, who attributed the worsening conditions (for example, managers being given 30 direct reports) to the narrative  of AI.  "Not, you'll note, the actual reality of AI, but rather, the story that AI lets you collapse the organization , slash headcount and salaries, and pauperize the (former) princes of labor. "The point of AI isn't to make workers more productive, it's to make them weaker when they bargain with their bosses. "Another of Bindley's sources went through eight rounds of interviews with a company, received an offer, countered with a request for 12% more than the offer, and had the job withdrawn, because the company didn’t want to move ahead anymore based on the way the compensation conversation had gone."

"Government agencies including the IRS, the FBI, DHS, and the Department of Defense have all purchased cellphone-location data, and possibly collected them too, via secretive groups such as the National Geospatial-Intelligence Agency.  "That means the government has at least some ability to map or re-create the past everyday movements of some American citizens. This is hardly even a cursory list of what is publicly known. "Advancements in artificial intelligence promise to turn this unwieldy mass of data and metadata into something easily searchable, politically weaponizable, and maybe even profitable. "DOGE is reportedly attempting to build a master database  of immigrant data to aid in deportations; NIH Director Jay Bhattacharya has floated the possibility of an autism registry (though the administration quickly walked it back).  "America already has all the technology it needs to build a draconian surveillance society —the conditions for such a dystopia have...

"What’s the future of any book publishing? Are most of the major New York publishing houses, as they dump their older editors, also weaning themselves of the literary in order to score bigger markets?   "Will there be books in fifty years?  "My son, in his mid-thirties, thinks not. A headline in The New York Times this morning announces A.I. Can Write Poetry, but It Struggles with Math .  "I’m no good with predictions. I’m devoted to something that has value to me, whether or not there is a place for it in our rapidly changing culture. I know others who feel the same way.  "Maybe the fact that poetry has, for the most part, always existed outside the market economy bodes well for its place in the future.  "In an interview, Raúl Zurita once said to me,  Because it is what opposes death, poetry is the hope of what has no hope. It is the possibility of what has absolutely no possibility. It is the love of what has no love. Like death, poetry was born with t...

"Given the hardware requirements, performance demands, and quality of the results, the stars really need to align for these newer Nvidia Broadcast features to feel truly worthwhile.   "If you have an Nvidia-powered system, by all means, play with the tool . Some of the features can come in handy, like the auto-framing one.  "But I wouldn’t recommend shelling out for a new Nvidia GPU just so you can save money on audio and video recording gear, especially if you want to get anywhere close to professional quality.  "And don’t forget that the power draw of the GPU trying to run these features will add up over time. "The audio quality I got from studio voice —perhaps limited by the RTX 4060 in my system —wasn’t something I’d want to share with any kind of audience on a regular basis, and it paled in comparison to the quality I could get just from having a headset with a boom mic.  "I’ve tested a lot of gaming headsets, and even very cheap wired headsets with a...

"In many ways, all is not well on American college campuses. Humanities enrollments are plummeting, and the academic job market for Ph.D.s has effectively collapsed. "These are grim times for the disciplines entrusted with carrying forward the humanistic project. "And yet, odd as it may seem, I think things have never looked better. Let the machines show us what can be done with analytic manipulation of the manifold.  "After all, what have we given them to work with? The archive. The total archive. And it turns out that one can do quite a lot with the archive. "In this sense, generative A.I. might count as a conceptual win for my field. Historians have long extolled the power of the archive . Little did we know that the engineers would come along and plug it in.  "And it turns out that a huge amount of what we seek from a human person can be simulated through this Frankensteinian reanimation of our collective dead letters . What a discovery!  "We hav...

"Microsoft has confirmed that Windows Recall is rolling out to everyone with Windows 11 KB5055627 on Copilot+ PCs.  "Recall is an AI-powered feature that captures your screen, so you can quickly find and jump back into what you have seen before on your PC.  "As Microsoft describes, you can use Recall to recall the moment you saw on your PC. "You Should Know: Windows Recall works by continuously capturing snapshots of your screen and using AI to index the content.  "Here are key technical details and commands to work with this feature. "

“We think that this is likely a very common occurrence in nature, where, many times, when you have simple cellular collectives, just through the very fact that they are interacting, it will lead to emergent physical traits,” Peter Yunker said. “These will very often have biological consequences.” Yunker’s work with the contact angle is a key example of how local factors can have macro consequences for life, said Ming Guo, who studies the geometry of developing embryos.  The results have implications for his own lab, he said, and for our understanding of how a group of individual cells becomes one multicellular individual. “This is going to be really interesting for us to further dive down and understand the rules that regulate how cells talk to their neighbors,” he said.   He hopes researchers can build on the work to show how these local communications lead to the emergence of a global shape.  With enough information, Guo said that he might one day be able to predict a...

"Perhaps if Io can be understood, so too can our moon —as well as several of the other satellites in our solar system with hidden tidal engines. "For now, this volcanic orb remains maddeningly inscrutable. Io’s a complicated beast,  Davies said. The more we observe it, the more sophisticated the data and the analyses, the more puzzling it becomes. "Our own moon shows evidence of primeval tidal heating, too. Its oldest crystals formed 4.51 billion years ago from the stream of molten matter that got blasted off Earth by a giant impact event.  "But a lot of lunar crystals seem to have formed from a second reservoir of molten rock 4.35 billion years ago. Where did that later magma come from? "Nimmo and co-authors offered one idea in a paper published in Nature in December: Maybe Earth’s moon was like Io. The moon was significantly closer to Earth back then, and the gravitational fields from the Earth and the sun were battling for control. At a certain threshold, ...

"China has signed nearly 200 intergovernmental space cooperation agreements with foreign countries and international organizations, covering fields from satellite development and lunar exploration to manned space flight, state media reported in December. "China’s space administration announced Thursday it has also approved loans of lunar samples collected by China’s earlier moon exploration mission to seven institutions across six countries, including Pakistan. "Quentin Parker, from the University of Hong Kong, emphasized the importance of international space cooperation: "The way that things are in the world at the moment is increasingly complex —but it’s important that this outreach and collaborative spirit is maintained,  he said." 

CEO Aravind Srinivas said this week on the TBPN podcast that one reason Perplexity is building its own browser is to collect data on everything users do outside of its own app.   This so it can sell premium ads. “That’s kind of one of the other reasons we wanted to build a browser, is we want to get data even outside the app to better understand you,” Srinivas said. “Because some of the prompts that people do in these AIs is purely work-related. It’s not like that’s personal .” And work-related queries won’t help the AI company build an accurate-enough dossier . “On the other hand, what are the things you’re buying; which hotels are you going [to]; which restaurants are you going to; what are you spending time browsing, tells us so much more about you,” he explained. Srinivas believes that Perplexity’s browser users will be fine with such tracking because the ads should be more relevant to them. 

"The thing the nonprofit board currently controls —governance of the world’s leading AI lab —makes no sense for the nonprofit to sell at any price.  "The nonprofit is supposed to act in pursuit of a highly specific mission: making AI go well for all of humanity. "But having the power to make rules for OpenAI is worth more than even a mind-bogglingly large sum of money for that mission. "Nonprofit control over how AGI is developed and governed is so important to OpenAI’s mission that removing control would violate the special fiduciary duty owed to the nonprofit’s beneficiaries,   the letter argues .  "Those beneficiaries are all of us, and the argument is that a big foundation has nothing on a role guiding OpenAI. "And it’s not just saying that the move is a bad thing.  "It’s saying that the board would be illegally breaching their duties if they went forward with it and the attorneys general of California and Delaware —to whom the letter is addresse...

"Rose Yu, associate professor at the University of California, San Diego (UCSD), is a leader in a field known as physics-guided deep learning , having spent years incorporating our knowledge of physics into artificial neural networks.  "The work has not only introduced novel techniques for building and training these systems, but it’s also allowed her to make progress on several real-world applications.  "She has drawn on principles of fluid dynamics to improve traffic predictions, sped up simulations of turbulence to enhance our understanding of hurricanes and devised tools that helped predict the spread of Covid-19. "This work has brought Yu closer to her grand dream —deploying a suite of digital lab assistants that she calls AI Scientist.  "She now envisions what she calls a partnership  between human researchers and AI tools, fully based on the tenets of physics, and thus capable of yielding new scientific insights.  "Combining inputs from a team of su...

Recently, a study published in Science Advances provided the strongest evidence yet that a brain microbiome can and does exist in healthy vertebrates — fish , specifically.  Researchers at the University of New Mexico discovered communities of bacteria thriving in salmon and trout brains .  Many of the microbial species have special adaptations that allow them to survive in brain tissue, as well as techniques to cross the protective blood-brain barrier. Matthew Olm, a physiologist who studies the human microbiome at the University of Colorado, Boulder and was not involved with the study, is inherently skeptical  of the idea that populations of microbes could live in the brain, he said.  But he found the new research convincing. “This is concrete evidence that brain microbiomes do exist in vertebrates,” he said. “And so the idea that humans have a brain microbiome is not outlandish .”

Researchers at Cybernews have uncovered a major privacy breach involving WorkComposer, a workplace surveillance app used by over 200,000 people across countless companies. The app, designed to track productivity by logging activity and snapping regular screenshots of employees’ screens, left over 21 million images exposed in an unsecured Amazon S3 bucket, broadcasting how workers go about their day frame by frame. The leaked data is extremely sensitive, as millions of screenshots from employees' devices could not only expose full-screen captures of emails, internal chats, and confidential business documents, but also contain login pages, credentials, API keys, and other sensitive information that could be exploited to attack businesses worldwide.

"Rivers do not drink their own water; trees do not eat their own fruit. The sun does not shine on itself and flowers do not spread their fragrance for themselves. Living for others is a rule of nature. We are all born to help each other."   —Pope Francis

“We want AI Jesus to feel like it’s the real Jesus talking to you ,” the artists told Creative Bloom.  “We fine-tuned his personality and speaking style in a way we hope is authentic . Then, we made sure that the (divine) product recommendations he makes are integrated into the conversation so they feel natural. We haven’t had any complaints… yet.” I mean, Jesus was the ultimate influencer back in His day, right? “His message spread across continents, and he built a following that outlasted empires. That kind of reach makes today’s influencers look like amateurs,” the creators told the outlet.  “But there’s also this beautiful tension: pulling someone from the distant past into the hyper-connected present.”

Health insurance giant Blue Shield of California is notifying millions of people of a data breach.  The company confirmed on Wednesday that it had been sharing patients’ private health information with tech and advertising giant Google since 2021. The insurer said that the data sharing stopped in January 2024, but it only learned this February that the years-long collection contained patients’ personal and sensitive health information. Blue Shield said it used Google Analytics to track how its customers used its websites, but a misconfiguration had allowed for personal and health information to be collected as well, such as the search terms that patients used on its website to find healthcare providers. The insurance giant said Google “may have used this data to conduct focused ad campaigns back to those individual members.” 

Nearly two months after hundreds of prospective California lawyers complained that their bar exams were plagued with technical problems and irregularities, the state’s legal licensing body has caused fresh outrage by admitting that some multiple-choice questions were developed with the aid of artificial intelligence. The State Bar of California said in a news release Monday that it will ask the California Supreme Court to adjust test scores for those who took its February bar exam. But it declined to acknowledge significant problems with its multiple-choice questions — even as it revealed that a subset of questions were recycled from a first-year law student exam , while others were developed with the assistance of AI by ACS Ventures, the State Bar’s independent psychometrician. “The debacle that was the February 2025 bar exam is worse than we imagined,” said Mary Basick, assistant dean of academic skills at UC Irvine Law School. “I’m almost speechless. Having the questions drafted by ...

"She is put in charge of hiring the new chief of China operations, giving her access to a voluminous paper-trail detailing the company's dealings with the Chinese government. "According to Wynn-Williams, Facebook actually built an extensive censorship and surveillance system for the Chinese state —spies, cops and military —to use against Chinese Facebook users, and FB users globally. "They promise to set up caches of global FB content in China that the Chinese state can use to monitor all Facebook activity, everywhere, with the implication that they'll be able to spy on private communications, and censor content for non-Chinese users. "Despite all of this, Facebook is never given access to China.  "However, the Chinese state is able to use the tools Facebook built for it to attack independence movements, the free press and dissident uprisings in Hong Kong and Taiwan."

"The world's economies are deeply, inexorably international, and the current trade war is going to disrupt things in ways we are not anticipating," he cautioned.  "And it's not going to be good," he wrote. "And when it gets better, it won't be in a world where the US is the dominant superpower." That bleak prognosis came with a sliver of hope. Schneier's final takeaway was the growing awareness of these complex issues within society and the cybersecurity community.  By openly addressing challenges and advocating for proactive, informed engagement, Schneier expressed confidence that collective action and intelligent solutions can ultimately enhance resilience and security. Schneier was hopeful the development of multiple AI models will lead to more diverse and distributed AI landscape. Development of diverse AI models is already a reality as the EU pushes for sovereignty in AI technology.

"On Tuesday, The Washington Post announced a new 'strategic partnership' with OpenAI.  "With this deal, ChatGPT will feature WaPo ‘s news content in generated responses to users’ searches.  "This will include 'summaries, quotes, and links to original reporting from The Post in response to relevant questions,' according to the outlet.  "'This partnership reflects a shared commitment to making reliable, factual information easier to find and engage with, especially on complex or fast-moving topics, where timely, well-sourced reporting, like that of The Post , matters most,'  WaPo explained in a press release.   "'ChatGPT will highlight The Post ’s journalism across politics, global affairs, business, technology, and more, always with clear attribution and direct links to full articles so people can explore topics in greater depth and context'."

Films made with the help of artificial intelligence (AI) will be able to win top awards at the Oscars, according to its organisers. The Academy of Motion Picture Arts and Sciences issued new rules on Monday which said the use of AI and other digital tools would "neither help nor harm the chances of achieving a nomination". Generative AI — which can create text, images, audio and video in response to simple text prompts —helped to produce some of the films awarded top industry accolades in March. But the Academy said it would still consider human involvement when selecting its winners.

"If you sometimes feel that the internet isn’t the same vibrant place it used to be, you’re not alone.  "New research suggests that most of the traffic traversing the network isn’t human at all. "Bots (software programs that interact with web sites) have been ubiquitous for years. But in its 2025 Bad Bot Report , application security company Imperva claimed this is the first time traffic from bots became more prevalent than human traffic. "The rise in bots is down to generative artificial intelligence (AI), Imperva said.  "This is the same technology that now flirts with people online for you and automatically writes heartfelt consolatory emails on behalf of heartless administrators.  "This tech has made it easier to create bots that do your bidding online. While some of those bots are benign, not all have your best interests at heart."

"In their modern incarnation, these ideas largely originate in science fiction novels. "That is to say, they were made up and popularized by people like me, the vast majority of whom made no pretense of being able to predict the future or even realistically describe a path from the present to the future they were presenting.  "Science fiction is something between a card trick and a consensual con game , where the writer shows you just enough detail to make you think that the rest of it must be lurking somewhere in the wings.  "No one in sf [science fiction] has ever explained how consciousness uploading could possibly work, and neither have any of the advocates for consciousness uploading —the difference is that (most of) the sf writers know they're just making stuff up."

The control and mechanics of rapid pigment translocation has been well studied in a number of different species, in particular amphibians and teleost fish. It has been demonstrated that the process can be under hormonal or neuronal control or both and for many species of bony fishes it is known that chromatophores can respond directly to environmental stimuli like visible light, UV-radiation, temperature, pH, chemicals, etc. Neurochemicals that are known to translocate pigment include noradrenaline, through its receptor on the surface on melanophores. The primary hormones involved in regulating translocation appear to be the melanocortins, melatonin, and melanin-concentrating hormone (MCH), that are produced mainly in the pituitary, pineal gland, and hypothalamus, respectively. These hormones may also be generated in a paracrine fashion by cells in the skin. At the surface of the melanophore, the hormones have been shown to activate specific G-protein-coupled receptors that, in turn, t...

🪷AI Overview Reproduction is the biological process by which new organisms are produced, while mimicry is an evolutionary adaptation where one organism (the mimic) resembles another organism or an environmental feature (the model) to gain an advantage.  Mimicry can involve various types of resemblance, like color, shape, behavior, or sounds, and often serves to deceive predators, prey, or mates.  Reproduction: Definition: The process of creating new individuals of a species.  Types: Includes sexual reproduction (requiring two parent organisms, like humans and animals) and asexual reproduction (requiring only one parent organism, like some plants and bacteria).  Goal: To continue the species and pass on genetic information to the next generation.  Examples: Fertilization in animals, budding in yeast, seed production in plants.  Mimicry: Definition: An evolutionary adaptation where one organism superficially resembles another organism or an environmental fea...

⭐AI Overview  A response is a reaction or reply to a stimulus or question, while a decision is a deliberate choice made after considering alternatives. The key difference lies in the level of deliberation and purpose. A response is often immediate and reflexive, whereas a decision is typically a more thoughtful and intentional act of selection.  Here's a more detailed breakdown: Response: Definition: A reaction or reply to a stimulus, question, or situation. Characteristics: Often immediate and automatic. May not involve extensive consideration of alternatives. Can be verbal, physical, or emotional. Examples: Saying "hello" in response to a greeting. Flinching when startled by a sudden noise. Answering a phone call.  Decision: Definition: A choice made after considering and evaluating different options or possibilities. Characteristics: Involves deliberation and weighing of options. Typically involves a goal or desired outcome. Can be personal, professional, or societal. ...

🐙AI Overview  The philosophy of mind explores the nature of mental phenomena, including consciousness, thought, and emotions, and their relationship to the physical world. It delves into questions like how the mind relates to the body, the nature of consciousness, and the existence of subjective experiences.  Here's a more detailed look: Key areas of inquiry in the philosophy of mind: Mind-Body Problem: This fundamental question explores the relationship between mental and physical entities. Does the mind exist independently of the body, or are mental states simply brain states?  Consciousness: Investigates the nature of subjective experience, including qualia (the "what it's like" aspect of experience) and the "hard problem" of explaining why conscious experience arises from physical processes.  Intentionality: Examines the property of directedness or "aboutness" in mental states, meaning they are about something in the world (e.g., a belief abou...

✨AI Overview  The first US census in 1790 was a relatively simple, paper-based process focused on basic demographic information, primarily for apportionment of congressional seats and taxation purposes. The last census in 2020, now digital, collects far more detailed data on a much larger population, encompassing a wide array of demographic, social, and economic characteristics.  Here's a more detailed contrast: First Census (1790): Purpose: Primarily to determine Congressional representation and, indirectly, to help with taxation.  Scope: Enumerated the population in the original 13 states, plus Kentucky, Maine, Vermont, and the Southwest Territory (Tennessee), according to Census.gov.  Method: U.S. marshals collected data using paper forms, with the names of heads of households and a general demographic count of remaining members.  Data Collection: Focused on basic demographics like age, sex, race (free white males, free white females, other free persons, and ...

David Ginty’s asking the same fundamental questions he set out to answer more than a decade ago:  Where do the various touch neurons go, what are their end structures, and how do they capture the richness of the physical realm?  “We’ve gotten a pretty good handle on who’s who in the skin and what their response properties are,” Ginty said.  But what about the heart, lungs, larynx, esophagus, stomach, intestines and kidneys? What are the neurons that make muscles ache and fatigue, or trigger migraines, or cause milk to flow in a mother’s breast when her baby suckles? Ginty also wants to know how all these neurons connect to the brain to generate perceptions:  How does pressure and vibration across millions of nerve endings become a hug? How do we feel wetness, slipperiness or elasticity? “Think about squeezing a balloon,” he said. “Presumably no one sensory neuron type is going to encode squeeziness.” His work has transformed our understanding not only of individual t...

Step by mechanical step, dozens of humanoid robots took to the streets of Beijing early Saturday, joining thousands of their flesh-and-blood counterparts in a world-first half-marathon showcasing China's drive to lead the global race in cutting-edge technology. The 21-kilometer event held in the Chinese capital's E-Town — a state-backed hub for high-tech manufacturing —is billed as a groundbreaking effort to test the limits of bipedal robots in real-world conditions. Around 20 teams from across China are taking part in the competition —with robots ranging from 75 to 180 centimeters tall and weighing up to 88 kilograms. Some are running autonomously, while others are guided remotely by engineers, with machines and humans running on separate tracks. Engineers said the goal was to test the performance and reliability of the androids —emphasizing that finishing the race , not winning it, was the main objective.

"Various theories for the neural basis of consciousness have been proposed, suggesting a diversity of neural signs and mechanisms .  "We ask to what extent this diversity is real, or whether many theories share the same basic ideas with a potential for convergence towards a more unified theory of the neural basis of consciousness.  "For that purpose, we review and compare the various neural signs, measures, and mechanisms proposed in the different theories.  "We demonstrate that different theories focus on neural signs and measures of distinct aspects of neural activity including stimulus-related, prestimulus, and resting state activity as well as on distinct features of consciousness.  "Therefore, the various mechanisms proposed in the different theories may, in part, complement each other.  "Together, we provide insight into the shared basis and convergences (and, in part, discrepancies) of the different theories of consciousness.  "We conclude that...

"In collaboration with American experimental composer Alvin Lucier, who passed away in 2021, scientists and artists created an art installation using cerebral organoids developed from the composer’s white blood cells. "Hooked up to transducers and actuators, these organoids created music by using electrical impulses to strike brass metal plates arranged throughout the installation. "The art installation, called Revivification , analyzes  The nature of living beyond death,  The essence of creativity, and  The persistence of memory. "Scientists have explored the idea of a hybrid consciousness that creates a shared reality between biological beings and artificial intelligence, or other ways to upload our consciousness to computers —if consciousness turns out to be purely computational which the jury is still out on, to say the least."

Alvin Lucier donated his blood the year before his death in 2021, and it was then reprogrammed into stem cells at Harvard Medical School.  Those stem cells were then transformed by the Revivification team into cerebral organoids: three-dimensional structures that resemble a developing human brain. “At a time when generative AI is calling into question human agency, this project explores the challenges of locating creativity and artistic originality,” the team says. “Like many things that are important to us, it’s an impossible concept to measure. Perhaps its value cannot be judged by scientific protocols, yet it remains something that we as humans should place great value in. “ Revivification is an attempt to shine light on the sometimes dark possibilities of extending a person’s presence beyond the seemed finality of death.”

In its technical report for o3 and o4-mini, OpenAI writes that “more research is needed” to understand why hallucinations are getting worse as it scales up reasoning models. O3 and o4-mini perform better in some areas, including tasks related to coding and math. But because they “make more claims overall,” they’re often led to make “more accurate claims as well as more inaccurate/hallucinated claims,” per the report. OpenAI found that o3 hallucinated in response to 33% of questions on PersonQA, the company’s in-house benchmark for measuring the accuracy of a model’s knowledge about people . That’s roughly double the hallucination rate of OpenAI’s previous reasoning models, o1 and o3-mini, which scored 16% and 14.8%, respectively. O4-mini did even worse on PersonQA —hallucinating 48% of the time. Third-party testing by Transluce, a nonprofit AI research lab, also found evidence that o3 has a tendency to make up actions it took in the process of arriving at answers. In one example, Tran...

"David Ginty, who heads the neurobiology department at Harvard Medical School, has been studying this quirky cast of characters —the sensory neurons of touch —for more than two decades, and has gotten to know them better than anyone else ever has.  "He has learned their electrical language and what forces excite them, and charted their intricate paths into the skin and up to the brain. And, through feats of genetic engineering and chemical labeling, he has produced the colorful portraits on his walls. "'David Ginty is the emperor of touch,' said Alexander Chesler, a sensory neuroscientist at the National Institutes of Health. "'You look at his publication list and you go, Oh my God ,' said David Hughes, a neuroanatomist at the University of Glasgow. 'He’s so massively productive, and all his papers are published in the very highest-impact journals.' "Beyond the technical breakthroughs and the discoveries fit for biology textbooks, it’s ...

"We study a novel language model architecture that is capable of scaling test-time computation by implicitly reasoning in latent space. "Our model works by iterating a recurrent block, thereby unrolling to arbitrary depth at test-time.  "This stands in contrast to mainstream reasoning models that scale up compute by producing more tokens.  "Unlike approaches based on chain-of-thought, our approach  Does not require any specialized training data,  Can work with small context windows, and  Can capture types of reasoning that are not easily represented in words.  "We scale a proof-of-concept model to 3.5 billion parameters and 800 billion tokens.  "We show that the resulting model can improve its performance on reasoning benchmarks, sometimes dramatically, up to a computation load equivalent to 50 billion parameters." 

"We argue that language space may not always be optimal for reasoning. For example, most word tokens are primarily for textual coherence and not essential for reasoning, while some critical tokens require complex planning and pose huge challenges to LLMs. "To explore the potential of LLM reasoning in an unrestricted latent space instead of using natural language, we introduce a new paradigm Coconut (Chain of Continuous Thought).  "We utilize the last hidden state of the LLM as a representation of the reasoning state (termed continuous thought ). Rather than decoding this into a word token, we feed it back to the LLM as the subsequent input embedding directly in the continuous space.  "Experiments show that Coconut can effectively augment the LLM on several reasoning tasks.  "This novel latent reasoning paradigm leads to emergent advanced reasoning patterns : the continuous thought can encode multiple alternative next reasoning steps, allowing the model to perf...

"Leading companies, such as OpenAI and Anthropic, are already heavily invested in existing LLM architectures. Redoing them to incorporate latent space reasoning would require heavy reengineering, so it’s unlikely they’ll adopt such techniques anytime soon. "[Luke] Zettlemoyer also cautions that latent space reasoning may have its own shortcomings.  "Ultimately, the data that LLMs train on is based on text, and the traditional approach has been extremely successful at finding patterns in it.  "LLMs can learn any kind of reasoning pattern, as long as it exists in texts —ensuring that the models reason in ways that humans do. Letting LLMs reason without using words could mean they’ll work in ways that aren’t amenable to human thinking. 'Moving into a continuous space could allow for all kinds of possibilities that aren’t actually going to be helpful,' Zettlemoyer said. "But even so, we now know it’s at least possible for models to work this way. Reasoning ...

"When a user queries an LLM, an algorithm breaks that input text into a sequence of tokens . "The model then converts each token into a string of numbers called an embedding , fodder for the underlying mathematical machinery. An input of 10 tokens results in 10 embeddings, for example.  "The transformer then processes these embeddings through its various components, called layers . Each layer feeds its results into the next layer, gradually connecting each embedding to every other embedding. The final layer puts all this information together to generate one final set of embeddings. The last embedding in this sequence is called a hidden state  — hidden because it’s not exposed to the outside world.   "This hidden state contains all the relevant information needed for the model to predict the most likely next token, or word, to follow the initial input sequence of tokens. "This is only the start of the process. This predicted token is added to the end of the ini...

American police departments near the United States-Mexico border are paying hundreds of thousands of dollars for an unproven and secretive technology that uses AI-generated online personas designed to interact with and collect intelligence on “college protesters,” “radicalized” political activists, and suspected drug and human traffickers, according to internal documents, contracts, and communications 404 Media obtained via public records requests. Massive Blue, the New York-based company that is selling police departments this technology, calls its product Overwatch , which it markets as an “AI-powered force multiplier for public safety” that “deploys lifelike virtual agents, which infiltrate and engage criminal networks across various channels.”  According to a presentation obtained by 404 Media , Massive Blue is offering cops these virtual personas that can be deployed across the internet with the express purpose of interacting with suspects over text messages and social media....

There’s a somewhat concerning new trend going viral: People are using ChatGPT to figure out the location shown in pictures. This week, OpenAI released its newest AI models, o3 and o4-mini, both of which can uniquely “reason” through uploaded images.  In practice, the models can crop, rotate, and zoom in on photos —even blurry and distorted ones —to thoroughly analyze them. These image-analyzing capabilities, paired with the models’ ability to search the web, make for a potent location-finding tool.  Users on X quickly discovered that o3, in particular, is quite good at deducing cities, landmarks, and even restaurants and bars from subtle visual clues.

"The model doesn’t deserve all the credit here; these outputs could not have been created without the work of human artists whose work was in the training data.  "But with just a few prompts, the model generated songs that few people would pick out as machine-made.  "A few could easily have been played at a party without raising objections, and I found two I genuinely loved, even as a lifelong musician and generally picky music person.  "But sounding real is not the same thing as sounding original. The songs did not feel driven by oddities or anomalies —certainly not on the level of Beethoven’s jump scare . Nor did they seem to bend genres or cover great leaps between themes.  "In my test, people sometimes struggled to decide whether a song was AI-generated or simply bad .  "Sanchez says people may wonder who is behind the music. But 'at the end of the day, however much AI component, however much human component, it’s going to be art,' he says. ...

"Both large language models (LLMs) and the human brain develop internal models of reality to make accurate predictions. Both typically prefer choices with the strongest track records.  [PDF] "However, when faced with a creative challenge, LLMs remain committed to high-probability options while humans can opt for unproven ones. This paper delves into one way of making unlikely events plausible — amplifying the anomaly. "The concept involves extrapolating viable consequences from an unlikely proposition. Rather than being treated as oddball events or one-offs , the anomaly permeates the creative work.  "Notably, novelty and appropriateness can be in tension with each other, with high utility coming at the cost of low novelty.  "Amplifying the anomaly aligns these competing demands. It enhances originality: the rarer a proposition and the more thoroughly it is worked out, the more unique and surprising the result. At the same time, the effectiveness and value of t...

"If you dip a paintbrush loaded with red ink into a glass jar of water, the ink will diffuse and swirl into the water seemingly at random, eventually yielding a pale pink liquid.  "Diffusion models simulate this process in reverse, reconstructing legible forms from randomness. "For a sense of how this works for images, picture a photo of an elephant. To train the model, you make a copy of the photo, adding a layer of random black-and-white static on top.  "Make a second copy and add a bit more, and so on hundreds of times until the last image is pure static, with no elephant in sight. For each image in between, a statistical model predicts how much of the image is noise and how much is really the elephant. It compares its guesses with the right answers and learns from its mistakes.  "Over millions of these examples, the model gets better at de-noising  the images and connecting these patterns to descriptions like male Borneo elephant in an open field .  "N...

"BitNet b1.58 2B4T doesn’t sweep the floor with rival 2 billion-parameter models, to be clear, but it seemingly holds its own. "According to the researchers’ testing, the model surpasses Meta’s Llama 3.2 1B, Google’s Gemma 3 1B, and Alibaba’s Qwen 2.5 1.5B on benchmarks including GSM8K (a collection of grade-school-level math problems) and PIQA (which tests physical commonsense reasoning skills). "Perhaps more impressively, BitNet b1.58 2B4T is speedier than other models of its size —in some cases, twice the speed —while using a fraction of the memory. "There is a catch, however. Achieving that performance requires using Microsoft’s custom framework, bitnet.cpp, which only works with certain hardware at the moment. Absent from the list of supported chips are GPUs, which dominate the AI infrastructure landscape. "That’s all to say that bitnets may hold promise , particularly for resource-constrained devices. But compatibility is —and will likely remain —a big st...

Last week Immigration and Customs Enforcement (ICE) paid contracting giant Palantir tens of millions of dollars to make modifications to a powerful ICE database and search tool to allow “complete target analysis of known populations” and to update the tool’s targeting and enforcement priorities, according to procurement records reviewed by 404 Media . The records show that Palantir is actively working on, and making updates to, the technical infrastructure underpinning the Trump administration’s mass deportation efforts. The news comes after ICE agents arrested a green card holding student at his interview to become a U.S. citizen… At the same time, Palantir is running adverts at U.S. colleges which say a moment of reckoning has arrived for the West. Our culture has fallen into shallow consumerism while abandoning national purpose. Too few in Silicon Valley have asked what ought to be built —and why. We did. 

CISA says the U.S. government has extended MITRE's funding to ensure no continuity issues with the critical Common Vulnerabilities and Exposures (CVE) program. "The CVE Program is invaluable to cyber community and a priority of CISA," the U.S. cybersecurity agency told BleepingComputer . "Last night, CISA executed the option period on the contract to ensure there will be no lapse in critical CVE services. We appreciate our partners' and stakeholders' patience." BleepingComputer has learned that the extension of the contract is for 11 months.

"I’d like to elaborate a bit more on our views on scrapers for you today. "Scrapers have been a thorn in the side of sysadmins for a very long time, but it’s particularly important as LLM scrapers seize the entire Internet to feed into expensive, inefficient machine learning models —ignoring the copyright (or copyleft, as it were) of the data as they go.  "The serious costs and widespread performance issues and outages caused by reckless scrapers has been on everyone’s mind in the sysadmin community as of late, and has been the subject of much discussion online. "Aside from the much-appreciated responses of incredulity towards LLM operators, and support and compassion for sysadmins from much of the community, a significant minority views this problem as less important than we believe it to be.  "Many of their arguments reduce to victim blaming— It’s not that hard to handle this volume of traffic; We should be optimized to better deal with it; We need more cachi...

As Venezuela’s economic crisis worsened and its currency became nearly worthless around 2018, educated Venezuelans signed up on AI-training and freelancing platforms to earn in U.S. dollars.  They formed up to 75% of the workforce at companies like Mighty AI and Scale AI in 2018. Remotasks even created a special program to attract Venezuelan workers.  They annotated all kinds of data to train AI tools, such as vision models, autonomous vehicles, and warehousing robots. They also moderated violent content and wrote articles to optimize websites for search. But with the rise of generative AI, such digital jobs have become scarce and poorly paid, workers and researchers told Rest of World . Without formal contracts, the workers have little choice but to find ways to compete with AI, or quit. “Every year, there have been fewer tasks,” said Fuentes, who moved to Colombia in 2019 from Cabimas, Venezuela. “If a task doesn’t come up, I’m just waiting, living with uncertainty. Days, we...

"The single most impactful investment I’ve seen AI teams make isn’t a fancy evaluation dashboard —it’s building a customized interface that lets anyone examine what their AI is actually doing .  "I emphasize customized because every domain has unique needs that off-the-shelf tools rarely address.  "When reviewing apartment leasing conversations, you need to see the full chat history and scheduling context. For real-estate queries, you need the property details and source documents right there. Even small UX decisions —like where to place metadata or which filters to expose —can make the difference between a tool people actually use and one they avoid. "I’ve watched teams struggle with generic labeling interfaces, hunting through multiple systems just to understand a single interaction.  "The friction adds up: clicking through to different systems to see context, copying error descriptions into separate tracking sheets, switching between tools to verify informat...

"We show that efficient pretraining techniques can produce useful models over corpora too large for easy manual inspection but too small for typical LLM approaches .  "We employ a novel date-attribution pipeline in order to obtain a temporally-segmented dataset of five 10-million-word slices.  "We train two corresponding five-model batteries over these corpus segments, efficient pretraining and Llama3-8B parameter efficiently finetuned. "We find that the pretrained models are faster to train than the fine-tuned baselines and that they better respect the historical divisions of our corpus.  "Emphasizing speed and precision over ahistorical comprehensiveness enables a number of novel approaches to hypothesis discovery and testing in our target fields.  "Taking up diachronic linguistics as a testbed, we show that our method enables the detection of a diverse set of phenomena, including en masse lexical change, non-lexical (grammatical and morphological) cha...