MIT's Alex Zhang on Recursive Language Models, 10,000-agent swarms and the future of harnesses
a1zhang · x · 2026-10-02
Latent.Space features a long interview with MIT PhD Alex Zhang, first author of Recursive Language Models (RLMs). Key points:
- Claude Code, Codex and Pi are basically the same: they are different harnesses around the same powerful models, and primitive wrapping leaves much capability on the table.
- RLM idea: let models treat their own prompts as objects in an external environment, use code execution for context offloading and recursive subagents to generalize across tasks.
- Expert vs brute force: a single expert agent can sometimes replace a trillion-token brute-force search, making harness design matter more than raw compute.
- OpenAI experiment: a 10,000-agent experiment producing 130B output tokens hints at where LLMs are heading.
- Academia's edge: freedom to take weird, ambitious research bets — Zhang follows star PhDs like Shunyu Yao and Jack Morris previously featured on the pod.
More from AGI Musings
- Stanford professor: AI takes students from problem to working prototype in 80 minutes — HariSubramonyam · 2026-11-19
- Berggruet Winner Thomas Nagel Dismisses AI Consciousness Talk as 'Just Ridiculous' — Dr_Atoosa · 2026-11-19
- Dwarkesh Claims China Is Only 12 Hours — Not 12 Months — Behind Frontier AI Labs — teortaxesTex · 2026-11-19
- Researcher calls METR/Anthropic '2x speedup' RSI analysis statistically dubious — jessi_cata · 2026-11-19
- Physician: AI competition now measured in weeks, GPT-6 versions 7 days apart — ramos_casals · 2026-11-19
- Why pro-handover rhetoric fails: people don't trust ASI's loyalty to labs — 1a3orn · 2026-11-19