Stanford's Kundaje blasts AI×Bio rigor gaps and NeurIPS review quality
On October 3, Anshul Kundaje, a Stanford professor of computational biology and deep learning, posted a series of messages leveling systematic criticism at research standards in the AI×Bio intersection and at the review quality of top ML conferences, drawing wide attention.
Confirmed
- Kundaje called out serious problems with the BioReason paper: its companion BioReason benchmark has fundamental flaws in task definition and information leakage in the prompts, and he explicitly recommended that these tasks no longer be used. That the paper ranked in the top 20 at NeurIPS that year, in his view, 'utterly reflects the review standards ML conferences apply to biology work — a complete joke.'
- He clarified that his criticism targets the BioReason paper itself, not BioReason-Pro; students in his class checked and believe BioReason-Pro 'very likely has other issues,' though this awaits further verification.
- He argued that so-called 'multimodal biological reasoning models' are misnamed, lacking basic controls, with early half-baked research making exaggerated hype-driven claims without controlled experiments — a trend the AI for Science field needs to watch out for.
- He advised newcomers moving from AI into biology: work with wet-lab people and systematically study experimental design, confounders, and controls, rather than stopping at knowing assays and lab protocols.
- He called for stricter scientific norms across the field, saying rigor needs to increase 5x and hype needs to decrease 100x.
Unconfirmed
- Whether BioReason-Pro has other issues remains to be verified; currently only his students' preliminary assessment exists.
Why it matters
- This is a direct, named criticism of AI4Science methods from a prominent scholar spanning computational biology and deep learning, touching three pain points: benchmark construction quality, missing control experiments, and the professionalism of top-conference reviews — of direct reference value for AI×Bio practitioners in research design, submission, and review.
2026-10-03 ~ 2026-10-03 · 11 related posts
Primary sources
- [source] Researcher slams NeurIPS reviews as BioReason top-20 pick called a farce — anshulkundaje · 2026-10-03
- [source] Stanford prof warns BioReason benchmark has task definition flaws and prompt leakage — anshulkundaje · 2026-10-03
- Stanford's Kundaje Flags BioReason Paper for Missing Basic Controls Amid Hype — anshulkundaje · 2026-10-03
- Stanford's Kundaje Slams 'Hypey' AI4Science Studies Published Without Basic Controls — anshulkundaje · 2026-10-03
- Stanford professor slams AI×Bio hype: half-baked studies lacking basic controls — anshulkundaje · 2026-10-03
- Stanford professor to AI folks entering bio: learn CONTROLS from wet-lab scientists — anshulkundaje · 2026-10-03
- Stanford's Anshul Kundaje: AI-bio newcomers must learn experimental design and controls — anshulkundaje · 2026-10-03
- Stanford professor Anshul Kundaje calls out sloppy rigor in AI x Bio research — anshulkundaje · 2026-10-03
- Stanford's Kundaje: model-in-the-loop biology needs different experimental controls — anshulkundaje · 2026-10-03
- [source] Stanford Professor Slams AI-for-Biology Work That Ignores Controls and Causality — anshulkundaje · 2026-10-03
- Stanford Professor: Don't Expect Causality to Emerge from Training on Billions of Cells — anshulkundaje · 2026-10-03