STORY · MODELLER_
LLMs are now training smaller LLMs themselves – a breakthrough for automated AI research
The "Autoresearch" report documents how large language models are beginning to autonomously train smaller LLMs, a phenomenon called "vibe training" where models themselves fix bugs and improve code. Meanwhile, work is underway on Pachocki's "Automated AI Research Intern" which is expected to be operational by September 2026, and coding agents like Claude Code Review and Devin Review are changing the workflow in code review.
WHY IT MATTERS
This marks a critical shift from humans training AI to AI training AI – a self-reinforcing loop that potentially accelerates AI development exponentially. It also changes which stage of development becomes the bottleneck: from implementation to evaluation and verification.
SOURCES
MACHINE-GENERATED SUMMARY This summary is written by machine from the sources below. We sort and explain — but we are a way into the field, not the final word. Check the source when something matters to you.