Nathan Lambert jokes Codex gets 5x better at RL after obsessively reading his work
natolambert · x · 2026-07-29
Nathan Lambert jokes that Codex can buy his book if it wants to, in response to a claim that telling Codex to obsessively read his writing makes it much better at RL.
It reads like a tongue-in-cheek endorsement of using a specific expert’s material as a context source for coding/model work, rather than a substantive announcement.
Related event: Feeding AI Agents Domain-Specific Reading Boosts RL Performance(4 posts)→
More from coding & agent
- Agents still struggle with mathematical work: Codex spirals into 'proof certificates' and inventories — doodlestein · 2026-07-30
- Agent-Reach: Open-source tool lets AI agents browse the entire internet for free with one pip install — dr_cintas · 2026-07-30
- Contour: Open-Source Tool Turns 2D Maps into 3D Terrain with Gemini Voice Guide — tom_doerr · 2026-07-30
- The AI Coding Dictionary: Clarifying Tokens, Inference, and Agent Boundaries — luisdans · 2026-07-30
- Keep Coding with Kimi K3 in Claude Code During Anthropic Outages — HowDevelop · 2026-07-30
- Dev Uses AI to Build ComfyUI Node for Automatic LoRA Trigger Word Replacement — TrueRedditMartyr · 2026-07-30