Akarsh Kumar on ASAL: using foundation models to search for artificial life
Machine Learning Street Talk · youtube · 2026-10-08
Machine Learning Street Talk shares a clip from its conversation with Akarsh Kumar (MIT, Sakana AI) on ASAL, a method for searching artificial life with foundation models. Instead of hard-coding a single simulation like Conway's Game of Life, the approach parameterizes a space of simulations and lets a foundation model evaluate the emergent complexity: "From this super, super simple world, you run this, you get insane amounts of complexity." The full conversation is available on the channel.
Related event: MIT and Sakana AI Researcher on Searching Artificial Life(2 posts)→
More from Research
- Tacit Knowledge and RSI: Why Human-Machine Friction Resists Automation — soumitrashukla9 · 2026-10-09
- Toby Ord Adds Gamma Parameter to His Swarm Scaling RSI Estimate — tobyordoxford · 2026-10-09
- Toby Ord's Swarm Scaling: 10,000 Agents Solved a Math Problem for $20M — tobyordoxford · 2026-10-09
- Mathematician Emily Riehl in Science: AI Has Solved Problems, Not Math — neurovium · 2026-10-09
- Reflex demo: catch, carry, hand over — code and paper open-sourced — DJiafei · 2026-10-09
- Reflex hits 85.7% catch success from RGB-D and chains catch-carry-handoff demos — DJiafei · 2026-10-09