Insitro's Koller: High-Quality Experimental Data Vastly Outweighs Massive Associational Data
anshulkundaje · x · 2026-08-06
Daphne Koller, founder and CEO of Insitro, recently argued that understanding human biology requires about 1000x the quantity of data we've collected. However, the academic field's current chase for "dataset size" might be a red herring.
Researchers emphasize that high-quality data capable of reliably answering mechanistic questions is the true scarce resource for understanding disease. A few bytes representing one clean experimental result can easily outweigh the causal information content of terabytes of associational data. Quantity without quality is irrelevant.
Related event: Insitro Founder Highlights Bottlenecks in Bio-AI(4 posts)→
More from AGI Musings
- antirez: Calling LLM Chain-of-Thought a Symbolic System Is the Worst Mystification in Science History — antirez · 2026-08-07
- Karpathy Called AI Agents 'Slop', Author Says Take-off Began with Opus 4.5 — scaling01 · 2026-08-07
- If APIs already work, why does the world need MCP? — hoenest_opinion · 2026-08-07
- Lloyds Bank Urged to Reveal Hidden Human Costs of Its £2B AI Savings — nordicinst · 2026-08-07
- Legibility Isn't Always Bad: Lessons from Amateur Radio, Personal Airplanes, and Cryptography — josh_wills · 2026-08-07
- Deep Dive: What Comes After Large Language Models? — bigdata · 2026-08-07