Why AI spending is piling into coding: verifiable tasks rule, says Orosz thread
bendee983 · x · 2026-11-13
Gergely Orosz notes that inside most companies the bulk of AI spend goes to coding (with quants at a few hedge funds also heavy users), and asks whether coding is uniquely suited to long-running agent tasks thanks to programmatic validation loops.
bendee983 argues mass AI usage will stay confined to verifiable tasks: coding, data analysis, possibly scientific and math research. These domains allow many runs and formal verification of outputs, and humans consume the final product—not the intermediate tokens—so generated code volume doesn't dictate perceived quality. In fields like creative writing, humans consume the actual AI-generated tokens, so removing them from the loop is impossible and high-volume, high-quality AI content can't scale there.
More from AGI Musings
- Analyst: Everyone Underestimates Non-AI R&D Acceleration — It'll All Be Ripping Anyway — teortaxesTex · 2026-11-13
- Sentience Institute validates a psychometric scale for 'substratism' — moral bias against AI — jacyanthis · 2026-11-13
- Anthropic's alignment work paying off as OpenAI falls behind, argues researcher — haider1 · 2026-11-13
- Claim: Senior researchers at OpenAI and Anthropic have confirmed contested study result — NathanpmYoung · 2026-11-13
- Investor tszzl: Chinese open-source models have earned huge global goodwill and soft power — tszzl · 2026-11-13
- 'We Have a Right to Compute': Critics Warn US AI Regulation Follows Nuclear Energy's Playbook — nptacek · 2026-11-13