AI for Science

A Reddit user says they may have discovered an unknown planet after using Claude Code, Anthropic's AI coding tool, to analyze astronomical data. The post has drawn significant attention on Reddit and Hacker News, with commenters debating both the plausibility of the discovery and the growing role of AI tools in independent scientific research. If verified, the find would highlight how accessible AI coding assistants are letting amateurs tackle data-heavy science projects.

Anthropic's Opus 5.5 model, running as autonomous research agents, has reportedly identified two candidate materials for room-temperature magnetic semiconductors, a long-sought class of materials that could enable spintronic devices without cryogenic cooling. The work, described by AI benchmark firm Vals AI, is drawing attention for showing AI agents contributing genuine materials-science discoveries rather than incremental analysis.

The Allen Institute for Artificial Intelligence says it will present its open research at COLM 2026, covering work that spans from open language models to AI agents designed for science. The announcement highlights Ai2's continued focus on building and releasing open tools and models that researchers can freely use, study and extend across scientific domains.

The Bulletin of the Atomic Scientists warns that artificial intelligence does not need to reach superintelligence or act out of malice to set off a nuclear conflict. Ordinary failures—miscalibration, false alarms, misinterpreted data, or humans over-relying on automated systems in minutes-long decision windows—could be enough. The argument reframes AI nuclear risk around mundane systemic errors rather than Hollywood-style rogue machines.

A Bulletin essay argues that alarmist claims about artificial intelligence wiping out humanity rest on weak science, and that this doom rhetoric actually benefits major AI companies. The piece says scare narratives justify concentrated corporate control, regulatory barriers, and heavy investment, letting dominant firms shape policy while the technology's real, present harms get less attention.

Researchers reportedly raced to complete a proof of the Unique Games conjecture before AI systems could get there first, according to a Quanta Magazine report. The story highlights growing anxiety among mathematicians that machine-assisted tools are closing in on major open problems in theoretical computer science, compressing timelines for human-only breakthroughs.

Ning Xu's entry won Nikon Instruments' Small World in Motion microscopy video competition, but it has been ruled ineligible after organizers determined it used generative AI. A contest spokesperson confirmed the disqualification to Gizmodo. The case is drawing attention as scientific imaging contests grapple with how to judge entries created with AI tools.

Al Pacino is drawing fresh attention over S1m0ne, his 2002 science fiction film about a computer-generated actress, which commentators say predicted today's AI-generated media with striking accuracy. The actor is also trending in Sweden, with readers revisiting the film's now-prescient take on synthetic celebrities and digital fame.

Playwright Alan Ayckbourn's latest work, The Trial of Romeo Oscar, stages a courtroom drama in which a robot stands trial for murder. The play is being discussed alongside commentary in the net.wars column 'What's in a name', which examines its take on robotics, naming, and science fiction themes.

Nikon's Small World in Motion photomicrography competition has a new winner after the original first-place entry was disqualified for using AI-generated content. The disqualification has drawn attention to how scientific imaging contests verify submissions and whether AI tools should be allowed in microscopy photography, with Ars Technica covering the change at the top of the results.

Physicist and science commentator Sabine Hossenfelder has published a new piece, shared via Science for the Wonder of Growth, laying out three reasons biologists say artificial intelligence systems are not conscious, at least not yet. The argument feeds an ongoing scientific debate about whether current AI models can have subjective experience or merely simulate intelligent behaviour.

The US Department of Energy, the National Institutes of Health, and the Biohub network have announced a partnership to build foundational datasets for predictive biological AI models. The collaboration aims to generate the high-quality biological data needed to train models capable of forecasting biological behavior, a major step for AI-driven life sciences research.

IBM is highlighting that artificial intelligence systems have reached a level of competence in physics that researchers find unsettling. The claim, made without supporting detail, points to AI tools increasingly solving or modelling physics problems once considered the preserve of human experts, prompting debate about how quickly machine capabilities are advancing in fundamental science.

Nikon's Small World in Motion photomicrography competition has changed winners after the original entry was disqualified for using AI. The new winner is Nguyen Nam Nhat of Vietnam, recognised for a video capturing a roundworm alongside a single-celled Dileptus predator. The case highlights growing scrutiny over AI manipulation in scientific imaging contests.

A discussion is underway about whether solving a difficult mathematics problem actually counts as advancing mathematical science. The argument being made is that the real value of mathematical research often lies in the discoveries made along the way, rather than in the final result alone. The topic touches on the role of computing and artificial intelligence in how mathematical problems are tackled and what that means for genuine scientific progress.

A claim circulating online compares the global big data industry's economic size to that of the Czech Republic, arguing that if mathematicians and researchers had access to comparable resources, science would advance much faster. The remark is framed as a call for greater investment in mathematics and fundamental research, tying current enthusiasm around AI and big data back to the discipline that underpins it.

The Chan Zuckerberg Biohub is directing $1.8 billion into AI-powered biology research, with an emphasis on establishing data standards before scaling up artificial intelligence applications. Commentators argue the initiative highlights that shared, well-structured datasets are the essential foundation for AI to deliver reliable scientific breakthroughs in life sciences.