ResearchAI 'virtual cell' predicts effective drugs for triple‑negative breast cancer before treatmentNature reports an AI model trained on 38 million protein measurements at Westlake University that identifies which drugs are likely to work in triple‑negative breast cancer patients before any dose is given.2 min read
ResearchRichard Socher’s Recursive Aims to Automate AI Research with Recursive Self‑ImprovementRichard Socher — founder of You.com and AIX Ventures — has launched Recursive, a company pursuing recursive self‑improvement: AI systems that automate the process of inventing and improving AI itself.5 min read
ResearchUniversity of Surrey study finds generative AI eases holiday planningA study published in the Journal of Travel Research and reported by Phys.org examined interviews with British adults who used generative AI tools to plan real trips.2 min read
Research25 Fields Medalists Warn AI Could Undermine Mathematical UnderstandingTwenty-five Fields Medal recipients, including Terence Tao and Peter Scholze, signed an open letter arguing that rapid AI-driven production of proofs risks divorcing mathematical discovery from human understanding.4 min read
ResearchOpen models in the AI ecosystem: roles, competition and risksOpen-weight and open-source models play multiple roles in today’s AI ecosystem: as research tools, commercial complements to closed frontier models, and vectors for competition between the U.S.6 min read
ResearchComparison of the Strengths and Weaknesses of Fable 5.1 and GPT-6 AstraThe Fable 5.1 and GPT-6 Astra models were compared, revealing their strengths and weaknesses; the author produced a detailed analysis and a few days ago shared a related diagram in video form.1 min read
ResearchAI progress outpaces understanding: generalization, agent limits, and open-source economicsRapid advances in building AI systems are not matched by understanding how they generalize or when their strategies fail.3 min read
ResearchResearchers: Large Language Models Shape Social Reality Through 'Algorithmic Socialization'Renáta Németh and Domonkos Sik argue that the central question about large language models (LLMs) has shifted from whether they are intelligent to how they act as communication partners that shape social norms and meanings.3 min read
ResearchToolGrad: answer‑first method speeds creation of tool‑use datasets with textual “gradients”ToolGrad is an answer-first framework that generates verified API call chains and then produces corresponding user prompts, using LLM-produced textual feedback to iteratively build workflows.5 min read
ResearchUniversity faculty use Claude mainly for curriculum, research and admin tasks, often building custom toolsAnthropic analyzed roughly 74,000 anonymized Claude.ai conversations from higher-education email accounts (May 22–June 2, 2025) and combined that with interviews of 22 Northeastern University faculty to map how professors employ AI.4 min read
ResearchAutomated Paired‑Prompt Evaluation Finds Claude Sonnet 4.5 Comparable on Political Even‑HandednessAnthropic reports a new automated, open‑source evaluation for political even‑handedness and applies it to its Claude models and several competitor models.5 min read
ResearchMining Genomes with AI to Find New Antimicrobials, Led by César de la FuenteCésar de la Fuente and his multidisciplinary laboratory apply deep learning to genome and protein databases to accelerate the search for antimicrobial molecules that could counter rising drug resistance.4 min read
