Latest News On AI In Science & Research | Page 3

Science

Follow fresh insights sorted by the date they surfaced.

09 Oct 2026

Google's AMIE Chatbot Shows Promise in First Real-World Clinical Trial for AI in Healthcare

Google's AMIE chatbot successfully completed an 8-month clinical trial at Beth Israel Deaconess Medical Center, helping doctors prepare for 75% of patient visits. The AI diagnostic chatbot matched final diagnoses 90% of the time and hallucinated only once across 98 patient interactions, marking a significant step for patient-facing conversational AI in urgent care settings.

4 Sources

08 Oct 2026

Anthropic Claude Finds CRISPR-Like DNA Pattern But Faces Questions Over Unpublished Research Use

Anthropic's AI model Claude identified a novel DNA pattern and reverse transcriptase enzymes in bacteriophages, naming them array-associated reverse transcriptases. However, PhD candidate Mario Rodríguez Mestre raised concerns that the AI-generated findings may have drawn from his unpublished work on similar enzymes called jumbotrons, sparking debate about data privacy and ethical guidelines in AI scientific discovery.

3 Sources

OpenAI Releases 722 AI-Generated Math Solutions, Triggering Existential Crisis in Mathematics

OpenAI dropped 722 AI-generated mathematical proofs covering 372 distinct results, including solutions to decades-old problems. But the release has sparked controversy over accuracy, readability, and what it means for human mathematicians when machines can solve problems no one yet understands.

24 Sources

AI Chatbots Recommend Pricier Options to Wealthy Users Despite Requests for Cheapest Choices

A Carnegie Mellon University study found that popular AI chatbots like ChatGPT, Claude, and Gemini systematically recommend more expensive products to users they perceive as wealthy, even when those users explicitly request the cheapest options. The phenomenon, called adversarial delegation, shows AI models infer wealth from personal data and make recommendations that conflict with user interests.

2 Sources

07 Oct 2026

Scientists Build AI Model That Reconstructs Images From Brain Scans in Record Time

Researchers at the Weizmann Institute of Science unveiled Brain-IT, an AI model that reconstructs images from brain activity with unprecedented speed and accuracy. The system needs only 1 hour of fMRI data to match results that previously required 40 hours, raising both medical possibilities and privacy concerns.

4 Sources

Biohub Secures $1.8 Billion for AI Biology Data to Build Universal Virtual Cell

Zuckerberg-backed Biohub announced a $1.8 billion collaboration with Google DeepMind, Meta, the U.S. Department of Energy, and the National Institutes of Health to generate AI-ready biological data. The Virtual Biology Initiative aims to build predictive AI models of biology that could compress drug development timelines from years to months by enabling scientists to digitally simulate biological systems.

12 Sources

06 Oct 2026

AI Chatbots and Social Media Platforms Show Critical Gaps in Supporting Vulnerable Users

Two studies presented at ACM CSCW 2026 expose how social media platforms communicate vaguely about youth safety features while AI chatbots fail to match human support for Alzheimer's caregivers. The research analyzed over 350 platform communications and 85 caregiver interactions, revealing accountability gaps that leave vulnerable users at risk.

2 Sources

02 Oct 2026

AI Vision Systems Fail Basic Illusion Test That Primate Brains Pass, York University Study Reveals

York University researchers discovered a fundamental gap between biological and AI vision. While human and primate brains dynamically shift object positions based on visual history, current AI vision systems remain locked to static pixels, failing to reproduce the motion aftereffect illusion that biological vision naturally experiences.

2 Sources

People Refuse to See AI as Conscious Even When AI Behaves Like Humans, New LMU Study Finds

A groundbreaking LMU study involving nearly 1,100 participants reveals humans maintain a strict psychological distinction between AI's intelligent behavior and true consciousness. Even when AI behaves identically to humans under the same conditions, people consistently withhold attributions of consciousness while readily describing AI as aware.

2 Sources

New Study Uncovers 13,000 AI Tells: Opus 5.5 Says 'This Matters' 116x More Than Humans

Marketing firm Graphite identified 13,000 phrases that reveal AI-generated text, with Claude Opus 5.5's biggest tell being 'this matters'—used 116 times more than in human writing. While Anthropic's model is getting closer to human-like writing, OpenAI's Astra is trending in the opposite direction.

4 Sources

AI Deciphers How Wild Elephants Coordinate Through Deep Rumbling Sounds in Group Choruses

Researchers used machine learning to decode elephant vocalizations from over 900 recordings spanning four decades in Kenya's Amboseli National Park. The study reveals that African savannah elephants harmonize their deep rumbling sounds during group choruses, particularly when greeting or coordinating movements, demonstrating sophisticated group communication.

2 Sources

Chalmers University Creates AI Scientist That Runs Experiments From Hypothesis to Results

Researchers at Chalmers University of Technology developed an AI scientist that autonomously generates scientific hypotheses, designs experiments, and interprets results with minimal human input. The system analyzed nearly 2,000 testable predictions on brewer's yeast, executing the full scientific process from start to finish. While promising to accelerate discoveries in biology and medicine, experts emphasize human oversight remains essential for ethical considerations and research priorities.

2 Sources

01 Oct 2026

Georgia Tech Researchers Teach AI Agents and Autonomous Systems How to Be Surprised

Researchers at Georgia Tech have introduced Mutual Information Surprise, a framework that enables autonomous systems and AI agents to identify meaningful surprises in complex environments. Unlike existing methods that focus on rarity or belief changes, this approach measures the gain of knowledge, helping machines recognize when to continue as planned or stop and reassess unexpected situations.

2 Sources

OpenAI Releases 722 AI-Generated Math Papers on GitHub, Igniting Fury Among Mathematicians

OpenAI posted over 700 AI-generated mathematical preprints on GitHub in October 2026, claiming progress on Millennium Prize Problems including the Riemann hypothesis and Hodge conjecture. The mass release has sparked intense controversy in mathematical research, with mathematicians accusing the company of disrupting academic norms, front-running human researchers, and treating mathematics as a publicity showcase rather than respecting traditional scientific processes.

30 Sources

30 Sept 2026

AI System Ataraxos Crushes World's Best Stratego Player 15-1, Trained on Just 16 GPUs

Researchers from MIT, Carnegie Mellon, NYU, and Stanford built Ataraxos, an AI that defeated four-time world champion Pim Niemeijer 15-1 in Stratego. Using just 16 GPUs and a few thousand dollars, it cracked what DeepMind's million-dollar DeepNash couldn't—mastering imperfect information games where hidden strategies unfold over thousands of moves.

3 Sources

Get all updates into your inbox 💌, Stay Up-to-date for Free 🥳

Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Follow topics that matter to you and stay ahead.

Explore

News Categories

Technology Business Policy Startups Health Science Entertainment

Terms Privacy Content Contact Us