# Latent Space interviews MIT's Alex Zhang on Recursive Language Models and harness design

> Latent Space published an interview with MIT's Alex Zhang covering Recursive Language Models, AI-written GPU kernels, and agent swarms, saying an RLM-based harness was first to roughly solve ARC-AGI-3 before OpenAI's Astra.

- **Topic**: Research
- **Published**: 2026-10-02T06:15:32.682Z
- **Canonical URL**: https://highsignal.sh/stories/latent-space-interviews-mits-alex-zhang-on-recursive-language-models-and-5768bc87

## Why It Matters

A specialist podcast spotlighting an academic researcher's bets on context offloading and persistent subagents — the kind of harness work Latent Space frames as an underused layer on top of increasingly capable models — arrives while the day's attention sits on frontier model launches from Gemini and Claude.

## Primary Sources & Citations

- [Academia is for Ambition — Alex Zhang, MIT](https://www.latent.space/p/rlm) — *Latent Space* (Reporting)

---

[← Back to Headlines](https://highsignal.sh/)