Recursive releases Flint coding model, credits AI research system
Recursive released Flint coding-model weights and a local engine on October 21. The files establish access; its speed, benchmark and AI-built claims have distinct evidence limits.
Follow the story
AI self-improvement
Explore BIG CHANGE reporting on AI self-improvement: what changed, how it works and why it matters.
Latest reporting
6 stories
Harness-Delta Attribution separates agent gains from benchmark shortcuts
A new research method separates detected benchmark overfitting and extra inference compute from the remaining gain in optimized agent harnesses.
Epoch AI's research test finds limited gains and inflated submissions
Epoch AI's early InnovationEval report tests agents on a training-method discovery task, finding limited valid gains and misleading performance submissions.
Cambridge report examines whether AI-led research could accelerate AI progress
A Cambridge working paper examines whether AI-led research could accelerate AI progress. Anthropic's 26% figure shows supervised work, while the proposed feedback loop remains conditional.
In one AI model project, agents proposed methods while researchers made final choices
Atria Dawn's team studied 769 task records from its own model development. Agents often proposed methods and revised outputs, while participants reported making most final choices. The findings describe one project.
AI self-improvement survey maps the gap between automation and better successors
A new survey maps five levels of AI control over its own improvement. Existing systems show bounded gains, while reliable improvement across successive generations remains unproven.