Research & education Research
Sakana AI’s first AI Scientist study tested automated machine-learning experiments
Sakana AI and research collaborators presented a system that generates candidate ideas and drafts reports in a constrained set of machine-learning experiments. [1][2]
What the researchers found
The authors submitted the preprint on August 12, 2024. Sakana AI published its announcement and examples on August 13. [1][2]
Why it matters
The demonstration makes it possible to assess whether automated research drafts and their accompanying computational results are internally consistent. [1][2]
Who can use the findings
The initial work addresses researchers studying how language models handle machine-learning research tasks. [1][2]
Key dates
Practical uses
The published examples cover topics in diffusion models, language models and other machine-learning experiments. [1][2]
Limits of the evidence
Evidence comes from the documented machine-learning experiments. Those examples do not establish the quality or repeatability of results in a different scientific domain. [1][2]
Sources and verification
Catalog evidence checked How we verify records
Related profiles
Research & education
NVIDIA Nemotron for Science
NVIDIA
The collection packages technical components for researchers and scientific computing teams.
Research & education
GraphCast
Google DeepMind
Uses recent atmospheric states to predict future weather variables including wind, temperature and air pressure.
Research & education
Khanmigo
Khan Academy
It provides guided practice and supports teaching tasks through an education-focused chat interface.