AI Briefing#ai-research
The latest AI and developer news about #ai-research, with the original source and a short summary.
The latest AI and developer news about #ai-research, with the original source and a short summary.
Jan von Plato used Claude to digest 1,009 pages of Gödel's 1940s workbooks, yielding 663 theorems where about 400 were verified as stated in Coq.
The 2026 cohort includes researchers in 38 countries working on health, inclusive systems, and computer hardware.
The study argues that defining slop by community rejection allows for measurable governance of AI-generated content.
The verification committee, led by David Gross, confirmed the AI-generated proof is valid but too complex for immediate human comprehension.
Anthropic reports that Claude Fable 5.2 generated complete specs, board layouts, and firmware for three devices that Diode Computers built with no circuit fixes.
Systematic evaluation across 16 trajectories shows WSP enables substantial localized gains but suffers from inconsistent cross-task generalization and validation reliability.
The platform now covers approximately 101k full texts from 2026 conferences, highlighting a shift toward LLM agents and multimodal models.
Anthropic analyzed 200,000 Claude sessions and stated that AI significantly accelerates the data analysis and writing stages of scientific research.
Researchers argue for maintaining multiple plausible AI-generated models to identify where predictions disagree and guide future experiments.
The proof, verified by Lean, establishes that quantum computers cannot achieve exponential speedups on random inputs without specific structural patterns.
The preliminary study used hundreds of GPUs for microscopy and produced a first structural model 34 hours after purification, though the biological function of the particles remains unknown.
Researchers at UC Berkeley and UCSF developed a streaming system that produces the first sound of a sentence within about one second of detecting a brain signal, compared to an eight-second delay for a full sentence in their earlier approach.
The funding supports Alzheimer's research, compute efficiency, and AI safety, alongside a $3 million grant for INSEAD's Thrive Lab.
A recent paper's claim that GPT-6 Astra can improve a 0.8B Qwen model by editing weights is dismissed as a task any competent human researcher could perform using steering vectors.
The summary introduces three major trends—expanding agent evaluation scope, structured learning, and refining data quality—along with 10 papers.
An internal Anthropic research model generated 318,484 protein designs, of which 17 lab-tested candidates met criteria for binding scorpion neurotoxins.
A new benchmark from Ai2 reveals that small prompt changes can shift LLM novelty judgment accuracy by over 50 percentage points, and simple baselines outperform expensive purpose-built systems.
OpenAI released AI-generated proofs for the local smoothing conjecture and a refutation of the Exponential Time Hypothesis, while deferring the release of physics results to external expert collaborations.
The study demonstrates that AI can produce formal proofs for software properties, a distinct capability from code generation that helps identify defects missed by conventional testing.
Nine frontier and near-frontier models use difficult psycholinguistic features more often than human writing, attributed to optimization for non-communication goals.
A new framework identifies scientifically plausible research directions that existing communities are unlikely to propose, recovering 2025 combinations at 35.9 times chance.
A study of 124 Harvard social science papers found that Opus 4.8 outperformed human experts as peer reviewers across statistical, formal, and qualitative work.
Working autonomously for 64 hours, Claude produced 17 experimentally validated binders against four scorpion neurotoxins, with all 211 designs released as open data.
The reliability of LLM agents stems from externalized structures rather than model weights.