AI research
Machine learning, language and vision models, evaluation, safety and reasoning.
The core of the observatory: new papers on machine learning, large language models, multimodal systems, reasoning, evaluation and AI safety, from labs, universities and the arXiv preprint server.
How to read these signals
Many AI papers are preprints that have not been peer reviewed, and many benchmark gains do not carry over to real use. Read the evaluation conditions, the baselines, the failure cases, and whether code or data is available.
Every entry in the live feed is labelled by source type: Lab report (reported by the research team), Preprint (not necessarily peer reviewed) or Research news (institutional coverage). The label describes where it came from, not whether it is correct.
See the latest AI research signals