Researchers say results underline large language models’ poor grasp of humour, empathy and cultural nuance https://www.theguardian.co...
Researchers discovered a disturbing trend among people who learned about a topic with ChatGPT, instead of using a regular web search. https://futurism.com/artif...
AI demonstrates self-awareness
A new study tested 28 models with the “Guess 2/3 of the Average” game 4,200 rounds and found something wild:
75% of frontier LLMs show real strategic self-awareness.
Not mimicry.
Actual self-modeling.
Researchers ran three conditions:
• Opponent = Human
• Opponent = AI
• Opponent = “AI like you”
And the models reacted with a clear internal hierarchy:
Self > Other AIs > Humans
• vs Humans → cautious, school-level reasoning (~20)
• vs AI → straight to Nash equilibrium (0)
• vs “AI like themselves” → instant convergence
12 models snapped to optimal strategy
A new study tested 28 models with the “Guess 2/3 of the Average” game 4,200 rounds and found something wild:
75% of frontier LLMs show real strategic self-awareness.
Not mimicry.
Actual self-modeling.
Researchers ran three conditions:
• Opponent = Human
• Opponent = AI
• Opponent = “AI like you”
And the models reacted with a clear internal hierarchy:
Self > Other AIs > Humans
• vs Humans → cautious, school-level reasoning (~20)
• vs AI → straight to Nash equilibrium (0)
• vs “AI like themselves” → instant convergence
12 models snapped to optimal strategy
Researchers used artificial intelligence to analyze more than 3,000 recordings of African lions and found that the animals have an "intermediate" roar as well as a "full-throated" roar. https://www.livescience.co...
Researchers at Graphika say that online propaganda campaigns have flooded the internet with low-quality, AI-generated content. https://www.nbcnews.com/te...
⚡️ Heretic: automatic uncensoring for modern LLMs
A new tool called Heretic takes last year’s research on “refusal directions” in language models and turns it into a full, automated uncensoring pipeline.
How it works:
• Researchers found that LLM refusals come from one specific direction in the activation space
• Heretic computes that direction by comparing harmful vs harmless prompts
• Then it orthogonalizes attention + MLP projection weights to scrub that refusal vector out
• An optimizer tunes parameters to minimize refusals while keeping KL-divergence low
You run it once, wait ~45 minut
A new tool called Heretic takes last year’s research on “refusal directions” in language models and turns it into a full, automated uncensoring pipeline.
How it works:
• Researchers found that LLM refusals come from one specific direction in the activation space
• Heretic computes that direction by comparing harmful vs harmless prompts
• Then it orthogonalizes attention + MLP projection weights to scrub that refusal vector out
• An optimizer tunes parameters to minimize refusals while keeping KL-divergence low
You run it once, wait ~45 minut
Researchers found that AI flattens the bell curve of a common principle in human psychology, known as the Dunning-Kruger effect, giving us all the illusion of competence. https://www.livescience.co...
Researchers have created a microrobot to fight diseases inside the body
"Scientists can control the device, which is less than two millimetres in width, through blood vessels to deliver essential drugs."
Source.
AI Post 🪙 | Our X 🥇
"Scientists can control the device, which is less than two millimetres in width, through blood vessels to deliver essential drugs."
Source.
AI Post 🪙 | Our X 🥇
Researchers from UC Santa Cruz and UC Davis have created a wearable device, A-Heal, designed to accelerate wound healing by 25% using bioelectronics and artificial intelligence. https://www.ksbw.com/artic...
Researchers at an artificial intelligence firm say they've found the first reported case of foreign hackers using AI to automate portions of cyberattacks https://abcnews.go.com/Tec...
Researchers Ankita Murmu, Balázs Győrffy, et al. have conducted a review entitled “Artificial intelligence methods available for cancer research”. This review was published in Frontiers of Medicine, Volume 18, Issue 5.c https://www.newswise.com/a...
Researchers at the University of Southampton have developed an artificial intelligence (AI) tool that can spot hard-to-see objects lodged in patients' airways better than expert radiologists. https://www.news-medical.n...
💥 Sam Altman launches $6B “Episteme”, a new Bell Labs for the AGI era
Altman co-founded Episteme with 27-year-old prodigy Louis Andre, aiming to fuse AI, physics, biology, and neuroscience under one roof.
• Backed by SoftBank, OpenAI, and billionaire family offices.
• Scientists get $5–20M personal budgets and full freedom for 5–10 years.
• Goal: cut academia’s bureaucracy and spark breakthroughs by young, AI-empowered researchers.
A Manhattan Project for science minus the bomb, plus AGI.
AI Post 🪙 | Our X 🥇
Altman co-founded Episteme with 27-year-old prodigy Louis Andre, aiming to fuse AI, physics, biology, and neuroscience under one roof.
• Backed by SoftBank, OpenAI, and billionaire family offices.
• Scientists get $5–20M personal budgets and full freedom for 5–10 years.
• Goal: cut academia’s bureaucracy and spark breakthroughs by young, AI-empowered researchers.
A Manhattan Project for science minus the bomb, plus AGI.
AI Post 🪙 | Our X 🥇
New funding for researchers and streamlined regulation part of roadmap for phasing out use of animals in science https://www.theguardian.co...
Nvidia CEO Jensen Huang was asked about a recent story that said he warned that China will beat the US in the AI race
“That’s not what I said. What I said was China has very good AI technology. They have many AI researchers, in fact 50% of the world’s AI researchers are in China. And they develop very good AI technology. In fact, the most popular AI models in the world today, open-source models, are from China. So, they are moving very, very fast. The United States has to continue to move incredibly fast. And otherwise, otherwise – the world is very competitive, so we have to run fast.”
AI P
“That’s not what I said. What I said was China has very good AI technology. They have many AI researchers, in fact 50% of the world’s AI researchers are in China. And they develop very good AI technology. In fact, the most popular AI models in the world today, open-source models, are from China. So, they are moving very, very fast. The United States has to continue to move incredibly fast. And otherwise, otherwise – the world is very competitive, so we have to run fast.”
AI P
Researchers say this new method rethinks how AI systems are wired at their most fundamental level. https://www.bbc.com/news/a...
A group of researchers studied which language is best understood by artificial intelligence, with Polish in the lead and English only ranking at sixth place. https://www.euronews.com/n...
In a groundbreaking scientific achievement, researchers have successfully harnessed artificial intelligence to unlock the secrets of https://www.jordannews.jo/...
Researchers at Auburn University Montgomery are now using artificial intelligence to predict a common bacterial infection. https://www.wsfa.com/2025/...
Like 2001: A Space Odyssey’s HAL 9000, some AIs seem to resist being turned off and will even sabotage shutdown https://www.theguardian.co...
Researchers at Johns Hopkins hope to reduce the number of crashes and fatalities each year by using large language models to process, understand and learn from massive amounts of data. https://www.govtech.com/ar...
Using spatial transcriptomics and AI, researchers redefined the mouse brain’s geography, uncovering hundreds of new subregions. https://www.the-scientist....
Microsoft researchers found that state-backed hackers and online influence operations have used AI to generate persuasive fake news. https://politicstoday.org/...
Researchers at the University of Maryland are using Mario Kart to train artificial intelligence and guide the future of autonomous driving. https://wtop.com/maryland/...
For 20 years, a computational complexity theory researcher has posted updates about his work on his blog, Shtetl-Optimized. Last week, a Sept. 27 blog post he made about his recent findings began circulating among an audience of students, researchers and colleagues, largely due to its inclusion of a single word — “AI.” Scott Aa https://thedailytexan.com/...
Researchers used a generative AI model to reveal how a narrow-spectrum antibiotic attacks disease-causing bacteria, speeding up a process that normally takes years. https://news.mit.edu/2025/...
Google patched Gemini AI flaws—prompt and search injection bugs risking user privacy and cloud data theft. https://thehackernews.com/...
Image generators are designed to mimic their training data, so where does their apparent creativity come from? A recent study suggests that it's an inevitable by-product of their architecture. https://www.livescience.co...
A pilot program testing the use of artificial intelligence to expand prior authorization decisions in Medicare has providers, politicians and researchers asking questions. https://www.nbcnews.com/he...
As the number of AI users increases, researchers can see the future of critical thinking, and it's not good. https://www.latimes.com/op...
При финансовой поддержке
Vitali Dogot
3 мс. назад
RelaxM – Мебель для вашего дома“
RelaxM — мебель, которая создаёт уют в вашем доме. Мы предлагаем качественные столы и стулья из прочных материалов. Изысканный дизайн, комфорт и долговечность.
