LLM Reasoning Is Just Extensive Search — Experiment Time Remains the Bottleneck
JFPuget · x · 2026-10-14
Data scientist aerdem4 (amplified by JFPuget) on what hasn't changed in DS: the bottleneck was never dev time but experiment time — and still is. He argues LLM reasoning is essentially extensive search surfacing the internet's average opinions; outside-the-box ideas remain a human job. He also credits his fast-POC habit (reusing archived projects) and precise Googling as still-valuable skills, while others have rotted.
More from AGI Musings
- There is no such thing as an AI drug, argues researcher Patrick Kidger — PatrickKidger · 2026-10-14
- Demis Hassabis: AI could cause 'major' incidents within 2 years without safeguards — TorturedPoet30 · 2026-10-14
- The obvious AI use in teaching: autogenerated, personalized exercises — Pseudomanifold · 2026-10-14
- A New Zealand fusion startup is the only one pursuing the levitating dipole approach — skdh · 2026-10-14
- Same agent, same task, judged twice: strict scoring yields under 20% of human baseline — YvesMulkers · 2026-10-14
- SaaS founder warns restaurant analogy breaks down as prompts replace software building — ayushtweetshere · 2026-10-14