science-ai
20 posts tagged science-ai.
- Contrastive Discovery: Why Science Agents Should Weigh Rival Explanations, Not Just Reject One
- FlowAlign treats alignment as target-space geometry, not just matching pairs
- Protein model steering still breaks at the decoder
- Scientist World Turns Research Papers Into Runnable Environments
- Project ProtoStar puts the boring data layer under science AI
- OpenAI’s science claims need verification, not applause
- OpenAI’s GPT-Rosalind plan puts life-science AI inside academic labs
- Reading 4,000 Comic Books at Once: What the PAGE Framework Actually Does
- Food Processing Is a Health Signal, Not Just a Nutrition Label Detail
- When One Input Has Many Right Answers: Bi-FORK and the Bifurcation Problem in ML Surrogates
- SciExam Tests AI Agents on Climate Modeling Without Giving Them the Answer
- Flow matching gives neural dissimilarity metrics one common frame
- Jev shows why science workflows need semantic evals, not just final-answer grading
- ScienceBuddy and the case for training the harness before the model
- CRISPR screens need learned experiment pickers, not bigger chatbots
- Codex as a lab scout for antimicrobial search
- The DOE's Genesis Initiative: A Federal Bet on Open Models
- AskChem Moves the Unit of Retrieval From Paper to Claim
- OpenAI’s 100,000-researcher ChatGPT push is an access story, not a discovery story
- GPT-5 Pro’s immunology win is a workflow story