Researcher on working with LLMs: 'both the model and I ran out of ideas'
burny_tech · x · 2026-10-01
Researcher nasqret shares first-hand experience using LLMs for research: failure usually shows up as stagnation, and when both he and the model run out of ideas, it's time to stop. Telltale signs include the model producing weird speculative conjectures or rewriting computations into something useless. He notes models can't solve every problem yet, but sometimes pushing further pays off with accidental discoveries.
More from Research
- Nature paper: 10M-cell immune multiome atlas from 1,108 FinnGen donors links disease variants to regulation — anshulkundaje · 2026-10-01
- Duplicate repost of the same Nature immune multiome atlas paper — anshulkundaje · 2026-10-01
- Incident Arena benchmark puts coding agents on call for incident response — amankhan · 2026-10-01
- RLTL;DR paper: self-generated feedback breaks Pass@128=0 barrier to 14-31% Pass@1 — burny_tech · 2026-10-01
- 1000×1000 collection releases 133 dynamical processes, from chaos to brains and climate — burny_tech · 2026-10-01
- GRASP quadrotor tight-formation paper wins IROS 2026 Best Student Paper — NikolaiMatni · 2026-10-01