Scientific discovery's bottleneck has shifted from ideas to data
Terence Tao argues that where hypothesis generation was once science's scarce resource, the bottleneck has shifted toward collecting and verifying against large datasets — with direct implications for how AI could accelerate research.
Added: 7/23/2026
Details
Core information and root causes
"I'm not sure nowadays that hypothesis generation is the bottleneck anymore... [modern science] collect[s] big data first, and then tr[ies] to get hypotheses from it." — Terence Tao, on The Dwarkesh Podcast
Tao traces a shift across the history of science: from theory-first idea generation (the classic scientific method) to data-first pattern discovery. That reversal changes what actually limits the pace of discovery today.