Español · The Impact of AI
Verified reporting, original sources and clear analysis of what changes for people.
2026-09-28 · UK edition · 113 reports across the archive
Translation is temporarily unavailable. This page currently shows the UK English original. Open the UK edition.
A newly listed preprint describes an open evaluation arena where models face other models in games with different kinds of uncertainty. It tests strategic behaviour, not whether an agent is safe to run a business or make decisions for people.
A Nature review maps how AI is moving from parameter tuning toward proposing experimental layouts, while highlighting trade-offs between computational optimisation, practical construction, interpretability and reliability.
Researchers describe Robin, a multi-agent system that generates hypotheses, proposes experiments, analyses results and revises its ideas, including work on candidate therapies for dry age-related macular degeneration.
A Nature paper presents Google's Co-Scientist, a Gemini-based multi-agent system that searches, critiques and refines scientific hypotheses before handing them to researchers for assessment and experimental work.
The AI Scientist pipeline automates idea generation, coding, experiments and manuscript drafting in machine-learning research, offering a concrete test of what parts of computational science can be delegated.
A Nature study examines how human expertise and machine systems can be combined to plan chemical synthesis, moving beyond a single-model recommendation toward a collective workflow.
A Nature analysis reports that scientists using AI tools can produce more highly cited work, while research topics may become more concentrated around areas where data and models are already strong.
A preprint comparing research-agent frameworks generated 37,802 ideas from shared literature and found evidence that automated systems can narrow exploration even while increasing output.
Microsoft Research introduced RetroChimera, a system for predicting how small molecules can be synthesised, addressing a bottleneck between computational molecule design and practical laboratory production.
A 2026 review surveys specialist scientific models, research assistants, agents and hybrid experimental systems, while examining technical, epistemic and institutional limits.
Anthropic says Claude helped identify an uncharacterised enzyme system with CRISPR-like repeats. It is a meaningful research lead, not yet a validated biotechnology breakthrough.