AI in Science News

Demis Hassabis calls AI 'ultimate tool' for science

At a Nobel Prize Outreach event, Hassabis described AI as the 'ultimate tool' and said 'we're in the foothills of the singularity'.

AI tackles decipherment of Linear A, a lost Minoan script

Linear A, the writing system of the Bronze Age Minoan civilization, has no bilingual text like the Rosetta Stone, making it undeciphered. Researchers are applying AI to analyze patterns and attempt to crack the language.

OpenAI: coding agents boost scientific computing

Field report from OpenAI shows scientists using AI coding agents to modernize scientific computing, accelerating software development and discovery in genomics. Agents handle routine maintenance, optimization, and complete redesigns while researchers define goals.

Closing the data loop in AI-driven drug discovery

Since the 1950s, the cost of developing new drugs has doubled roughly every nine years, with failure rates above 90% and costs reaching $1B–$2.5B. AI is shifting drug discovery from empirical screening to predictive design, but requires robust data and lab integration.

NVIDIA advances semiconductor materials engineering and manufacturing

NVIDIA highlights how its CUDA-X and simulation platforms address growing compute demands in semiconductor manufacturing. The post covers materials engineering, digital twins, and computational chemistry to accelerate innovation.

Terence Tao: Mathematics in the Age of AI

Slide deck from Terence Tao's 2026 ICM talk examining how AI is transforming mathematical research, highlighting recent AI breakthroughs and their implications.

Tutorial explores FAIRChem v2 UMA for multidomain atomistic simulation

Covers FAIRChem v2 UMA, a universal machine-learning interatomic potential for molecules, catalysts, materials, vibrations, and molecular dynamics. Includes environment setup and Hugging Face authentication for the gated model.

Nvidia's new DNA model learns what token prediction misses

Nvidia introduces a new approach for DNA modeling that moves beyond token prediction, addressing limitations of text-generation models for structured genomics data. The model is designed to capture latent representations more effectively.

Nvidia is sending GPUs to the Moon

Nvidia announced it will send GPUs to the Moon as part of a new space initiative. The GPUs will enable AI processing capabilities in the lunar environment.

AI cracks century-old Jacobian conjecture

Anthropic's Claude Fable 5 solved the 87-year-old Jacobian conjecture, announced by Levant Alpöge. The result has been verified, sparking mixed reactions among mathematicians.

Google Research introduces quantum error-learning architecture

Google researchers developed a method for quantum computers to learn from errors, aiming to improve fault tolerance. The approach uses machine intelligence to adaptively correct noise in quantum circuits.

Terrence Tao uses ChatGPT to discuss Jacobian Conjecture counterexample

Fields Medalist Terrence Tao shared a ChatGPT conversation exploring a potential counterexample to the Jacobian Conjecture, a long-standing open problem in mathematics.

GPT-5.5 demonstrates problem-solving in pure functional analysis

A Reddit post shows GPT-5.5 solving selected problems in pure functional analysis, highlighting advanced mathematical reasoning capabilities.

Leiden Declaration shared by Emily M. Bender

AI agents improve Terence Tao's Collatz theorem bound

AI agents helped prove that for any function f(N)→∞, almost all N reach below f(N) in at most 436 ln N steps. The result is Lean-verified and establishes natural density, but does not prove the full conjecture.

Ai2 updates Asta AI agents for science with paper search and data handoff

AlphaFold applied in protein engineering

Tiny memristor chip cuts brain modeling time to under 10 milliseconds

A tiny memristor chip cuts brain modeling time to under 10 milliseconds. The chip is designed to accelerate computational brain models.

Human mathematicians are being outcounterexampled

A blog post discusses how AI models now generate counterexamples that human mathematicians struggle to produce, suggesting a shift in mathematical discovery.

Apply for Anthropic's AI for Science rare disease research grants

Researchers can receive up to $50,000 in Claude credits over six months. Two tracks: one for basic research and one for early-stage biotechs accelerating clinical development for rare diseases.

Amodei: AI diagnoses medical cases that stumped specialists

Anthropic CEO Dario Amodei, a former biologist, recounted cases where AI diagnosed complex medical issues that puzzled specialists. He argues that AI's most transformative impact is currently in biology and medicine labs.

Bezos backs CuspAI as startup teams up with Nvidia to hunt for chipmaking materials

Jeff Bezos invests in CuspAI, an AI startup partnered with Nvidia to discover new chipmaking materials. The company is part of a new wave of AI firms tackling physical-world challenges.

Claude Fable 5 produces counterexample to 87-year-old Jacobian conjecture

The model produced a hand-checkable counterexample to the Jacobian conjecture (1939), an open problem on Smale's list of 18 mathematical problems for the 21st century. Terrence Tao discussed the result in a ChatGPT conversation.

Talk explores engineering challenges of single-cell biology foundation models

Akram Baharlouei (Altos Labs) discusses building foundation models for single-cell biology from an ML engineering perspective. The talk covers data scaling, pretraining, and domain adaptation challenges.

Extends AI agents with 148 ready-to-use scientific research skills

Sina Shahandeh presents on autonomous agents for scientific tasks

The talk contrasts common Autoresearch tasks (coding puzzles, toy optimization) with the need for real measurement data in scientific discovery. Sina Shahandeh from Radicait discusses the challenges and requirements for autonomous agents to assist in genuine scientific research.

Weather Lab gets major update with AI weather models

AI reads ancient text buried under volcano for 2,000 years

Agentic tree search automates scientific manuscript writing

AI agent Biomni accelerates biomedical research

Biomni performs research tasks across diverse biomedical fields. The AI agent, described in Nature Medicine, could be a powerful research partner for scientists.

Databricks and Dotmatics launch AI-ready science platform

The platform integrates Databricks' data lakehouse with Dotmatics Luma to bridge the gap between experimental data and AI-driven insights. It aims to accelerate scientific discovery in healthcare and life sciences by making data AI-ready.

Podcast explores Lila Sciences' vision for AI-driven automated labs

Lila Sciences envisions fully automated labs where AI-guided robotics run experiments 24/7, likening them to dark warehouses of data centers. Founders Andy Beam and Rafa Gómez-Bombarelli discuss the convergence of AI and robotics to accelerate scientific discovery.

Two papers propose active learning for bioacoustic event classification to reduce…

Greedy Volume Maximization selects diverse gradient embeddings, while Adaptive Diversity-Uncertainty balances diversity and uncertainty with redundancy control. Both methods aim to reduce expert annotation in large-scale bioacoustic monitoring.

Documentary series chronicles Demis Hassabis's path from chess prodigy to Nobel Prize…

The 9-episode series 'The Nobel Call' traces Hassabis's 30-year journey to winning the 2024 Nobel Prize in Chemistry. Episode 1 covers his early life as a child chess master and his dream of building general AI.

DeepMind essay: AI agents face 'validation bottleneck' in science

Apple FaceID co-inventor launches Hemispheric for AI brain scans

Founded by Gidi Littwin, Hemispheric aims to make diagnostic brain scans using AI as cheap and easy as a blood test. The technology targets conditions like depression, PTSD, and Parkinson's.

ChatGPT proves 50-year-old math conjecture

ChatGPT successfully proved a mathematical conjecture that had remained unsolved for 50 years, according to a Scientific American report.

WikiSTAR system analyzes scientific Wikipedia article revisions

WikiSTAR uses NLP to surface scientifically meaningful revisions from Wikipedia's revision history. The system aims to reveal how scientific knowledge evolves on the platform.

Solving 20 Erdős problems with 20 parallel Codex accounts

A project solved 20 Erdős problems by running 20 Codex accounts in parallel, showcasing AI's potential in mathematical research.

Claude Code equipped with 199 bioinformatics skills; BixBench 92%

Tachikawa reports Claude Fable solved 6-month physics problem

Theoretical physicist Yuji Tachikawa of the University of Tokyo reported that Claude Fable solved a quantum field theory problem his group had been stuck on for six months. He describes the model as ushering in a new era of collaborative science.

Scientists use AI and quantum computing to generate new peptides

Researchers cobbled together funding to show how quantum computing and AI can generate novel peptides for drug development. The work targets underserved populations and rare diseases.

AI boosts scientific productivity but reduces creativity, study finds

AI models assess agricultural supply chain resilience

The tool integrates GTAP economic models with APSIM biophysical models to analyze supply chain disruptions. It enables simulation of shocks like climate events or market changes, providing insights for policymakers.

OpenCap Monocular estimates 3D human movement from single smartphone video

AI-generated videos designed to stimulate specific brain regions

Researchers at EPFL developed a method to generate AI videos optimized to drive activity in targeted brain regions. The project, NeVo, uses generative models to produce visual stimuli that maximally activate specific neural populations.

Crys-JEPA uses Joint Embedding Predictive Architecture for crystal learning

Microsoft unveils Aurora 1.5 open weather foundation model

Aurora 1.5 adds 22 weather variables and hourly resolution for applications in energy, agriculture, and climate risk. The model is released as an open foundation model with probabilistic ensemble forecasting.

LTX spins off from Lightricks as open world models company

LTX is spinning off from Lightricks to become an independent open world models company. It will release open foundation models for simulating physical reality, including how objects move and forces interact. The models remain fully open for enterprise use.

Claude Code rebuilds $7,500/yr structural biology suite, now free

A developer used Claude Code to rebuild functionality from a $7,500/year structural-biology suite and released it for free. The project addresses the gap between AI protein structure prediction and accessible software for wet-lab researchers.

Powering scientific discovery: BYOKG and GraphRAG for intelligent pharmaceutical research

The AWS AI Blog walks through using BYOKG and GraphRAG with Amazon Bedrock and Neptune to unify scattered pharmaceutical data for more efficient drug discovery. The technique integrates internal lab notes, published literature, and genomics databases.

OmniScientist automates scientific discovery with LLM agents

Claude Fable, meaningless if you work in science

Reddit user reports that even simple tasks involving scientific, clinical, or pharmacological keywords fail due to heavy safeguards, rendering the desktop automation agent unusable for life science research.

British startup launches longevity research lab into orbit

The orbital lab will transmit data to train AI models for predicting protein behavior linked to Alzheimer’s and cancer. This mission aims to leverage microgravity to accelerate biological research that is difficult to conduct on Earth.

Building a scaffold-split random forest QSAR co-scientist

Tutorial walks through building an AI co-scientist using ChEMBL, RDKit, SHAP, and BRICS for discovering EGFR inhibitors targeting the C797S osimertinib-resistant mutation in non-small cell lung cancer. Covers scaffold-splitting, random forest QSAR modeling, and explainable AI with SHAP.

OpenAI podcast explores AI-powered biology with Codex

Immunologist Derya Unutmaz discusses using OpenAI's Codex for cell analysis and simulating immune cells. The conversation covers the potential for AI to help scientists simulate experiments.

Synthetic Sciences releases OpenScience, an open-source AI workbench

OpenScience is licensed under Apache 2.0, runs on your own infrastructure, and is positioned as an open alternative to Anthropic's Claude Science. It is model-agnostic and supports research in machine learning, biology, physics, and chemistry.

AI-Researcher automates scientific literature review and manuscript creation

Damo Academy AI agent discovers four new superconductors

Alibaba's Damo Academy developed an AI agent called Elements Claw that discovered four new superconductors, potentially revolutionizing materials science. The AI agent autonomously searches for and identifies superconducting materials with high transition temperatures.

Multi-agent framework transforms research stories into scientific manuscripts

Composable scientific skills for rigorous AI research agent workflows

COrigami: end-to-end pipeline for co-design unveiled

NVIDIA launches BioNeMo Agent Toolkit for AI-driven biology and chemistry

The toolkit enables AI agents to perform hands-on scientific work using real scientific software, moving beyond conversational Q&A. It targets drug discovery and molecular research with agentic capabilities.

80TB astrophysical dataset released on Hugging Face

Drug Discovery In The Age of AI

Explores AI applications in pre-clinical drug development, highlighting how machine learning models accelerate target identification and lead optimization. Covers challenges in data quality and model interpretability.

Proto: a domain-specific programming language for generative biology

Proto is the first programming language designed specifically for generative biology. It aims to accelerate research in smart medicine and cancer therapies.

BoltzGen accelerates protein design on Amazon SageMaker AI

BoltzGen is a diffusion-based generative model for designing protein binders to specific targets. Amazon SageMaker AI manages end-to-end GPU infrastructure to accelerate design campaigns.

The moment Anthropic convinced me it's serious about science

In a STAT newsletter, reporter Brittany Trang recounts the experience that convinced her Anthropic is serious about AI in science. The piece highlights Claude's potential in healthcare and life sciences.

Diffusion research for molecular AI with Genesis founders on Latent Space

Latent Space podcast interviews Evan Feinberg and Sergey Edunov about diffusion models for molecular generation. The discussion covers their work at Genesis Molecular AI, a startup applying these techniques to drug discovery.

AI drone reforestation restores degraded Brazilian cattle farm

Terence Tao discusses AI in math via SAIR competitions

Terence Tao talks about how SAIR's mathematics competitions are pointing toward new research infrastructure for AI era. AI is accelerating proof generation and verification, but traditional systems weren't designed for AI-assisted proofs.

DEFRA uses AI and Databricks to accelerate peatland restoration

The UK's Department for Environment, Food and Rural Affairs (DEFRA) is using Databricks and AI to accelerate peatland mapping and restoration across England. The project, in collaboration with Natural England, aims to restore a critical ecosystem for carbon storage.

NVIDIA BioNeMo Agent Toolkit integrates with Anthropic Claude Science

The toolkit packages NVIDIA-accelerated models and libraries as callable skills within Claude Science, enabling natural-language-driven research workflows. 18 of the top 20 pharmaceutical companies already use NVIDIA BioNeMo.

Grant Sanderson – AI and the future of math

Grant Sanderson argues AI is making faster progress in math than other fields, revealing what AI progress will look like elsewhere. He also discusses conceptual breakthroughs versus current AI abilities and the overhang from connecting existing ideas.

Stanford HAI posts talks from 2026 Physics and AI Conference

Recordings feature speakers from Stanford, DeepMind, Anthropic on using foundation models and agents for physics. Conference held June 10-12, 2026.

Anthropic launches Claude Science, an AI workbench for scientists

Claude Science provides 60+ scientific databases and integrates with NVIDIA BioNeMo Agent Toolkit for accelerated life sciences research. Anthropic also announced an internal AI drug discovery program using the platform.

OpenAI launches GeneBench-Pro benchmark for genomics AI

GeneBench-Pro tests AI agents on messy biological data, analysis path selection, and real-world research judgment calls. The benchmark aims to measure progress in scientific reasoning beyond standard benchmarks.

SandboxAQ brings scientific AI models to Google Cloud Marketplace

First model available is AQCat for materials and catalyst discovery, built on real-world lab data and scientific equations. AQPotency for drug discovery will follow.