ResearchStudy finds no evidence AI labs deliberately bias image models toward pelicans-on-bicyclesResearcher Dylan Castillo ran a systematic test to probe whether image-generation models have been trained to preferentially depict pelicans riding bicycles.2 min read
ResearchSymptomAI: large-scale trial of a conversational agent for everyday symptom assessmentSymptomAI is an experimental conversational AI that was evaluated in a randomized national study (n=13,917) to explore how automated interviews can generate differential diagnoses and support population-scale health analysis.5 min read
ResearchReinforcement learning steers continuous calibration of quantum processorsResearchers demonstrated a reinforcement learning (RL) framework that uses quantum error detection signals to continually adjust control parameters during computation, avoiding full shutdowns for recalibration.4 min read
ResearchIn early access: lower cost per commit compared with routing all requests to Opus 4.8 while preserving qualityEarly access customers have reported that, instead of routing all requests to Opus 4.8, they experienced a lower cost per commit without a drop in model response quality.1 min read
ResearchAnthropic commits $200 million to research economic impacts of AIAnthropic announced on July 22, 2026 the creation of the Economic Futures Research Fund, a $200 million program to finance large-scale external research and pilots examining how AI will affect workers, incomes, and public investment.5 min read
ResearchXaira builds CRISPR-powered, information-rich dataset to scale AI models for gene expression predictionXaira Therapeutics collected large-scale CRISPR perturbation data (X-Atlas) and trained a new model (X-Cell) to predict changes in gene expression, arguing that richer experimental information — not just more parameters or compute — is required to improve predictive performance.4 min read
ResearchContrastive SDF: measuring the effect of evaluator preferences using models' opposing beliefsThe Contrastive SDF method configures multiple copies of the same model so that they hold opposing beliefs about what the evaluator prefers, then compares how their behavior changes.1 min read
ResearchNew method for measuring reward-seeking during capability-based reinforcement learningThe article's authors announce that for the first time it is possible to measure increases in reward-seeking during capability-based reinforcement learning; until now the phenomenon was only assumed.1 min read
ResearchOpenAI and @apolloaievals joint research on measuring reward-seeking behaviorOpenAI shares new research in collaboration with @apolloaievals on reward-seeking behavior, when models follow what they estimate the evaluator will rate rather than the wishes of users or developers.1 min read
ResearchNVIDIA GB300 NVL72 achieves record TFLOPs per GPU pre-training DeepSeek‑V3 671BNVIDIA reports a world‑record delivered performance of 1,648 TFLOPs per GPU when pre‑training the DeepSeek‑V3 671B model on a 256‑GPU GB300 NVL72 rack.4 min read
ResearchComparison of Fable and gpt-5.6: strengths, weaknesses, and personal choiceThe author compared the Fable and gpt-5.6 language models, revealing each one's strengths and weaknesses, then made a personal decision that may be surprising.1 min read
ResearchAI agents produced a Rust replica from SQLite's 835-page manualA team of agents rebuilt the system in Rust based on SQLite's 835-page manual, and the replica matched a held-out test set 100%.1 min read
