Initialization matters far more than expected in LLM-driven scientific discovery, new study finds
BenBlaiszik · x · 2026-10-15
New research from UChicago, Vector Institute and University of Toronto (Mansi Sakarvadia, Marco Ciccone, Colin Raffel) studies LLM-driven discovery, where models iteratively optimize objectives inside a "harness" to find algorithms, theorems, or drugs.
- Benchmarks five tasks (Can't Be Late, CloudCast, TSP, Circle Packing, prompt optimization), comparing initialized vs uninitialized modular populations
- Initialized runs converge faster and reach higher final scores across Gemini 3.5 Flash, GPT-OSS 120B, and Llama 3.1 Instruct 8B
- Conclusion: how you initialize a discovery harness matters far more than expected, boosting both token efficiency and ceiling
Related event: Initialization Key to LLM-Driven Scientific Discovery, Study Finds(3 posts)→
More from Research
- 167,356 simulated shop-days: paper shows eval world design writes the model behavior it scores — liminal_bardo · 2026-10-15
- Multilingual NLP roadmap: researcher maps open problems from data selection to LID across 350+ languages — StellaLisy · 2026-10-15
- Beyond RAG: Injecting Tensor Memory Cartridges Into Layer-3 of a Frozen LLM — Nearby_Indication474 · 2026-10-15
- Cohere Labs: more coding agents isn't a scaling strategy — when to delegate work — pvncher · 2026-10-15
- UCLA paper: productivity tech like AI can trigger a workplace "rat race" that hurts welfare — soumitrashukla9 · 2026-10-15
- Deterministic Miller-Rabin primality testing now in P after last week's breakthrough, ~O((log n)^9.9984) — geoffreyirving · 2026-10-15