Joules-to-GDP framework measures economic value AI creates per unit of energy
karinanguyen · x · 2026-11-07
Thoughtful's team introduces Joules-to-GDP (J2G), a framework measuring how efficiently AI converts energy into economic value, and which tasks, models, and hardware give the best return. It builds on OpenAI's GDPval across 44 occupations in finance, engineering, law, and admin work.
Key findings:
- Highest-value tasks ≠ highest value per joule: CAD design earns 10% more per task than building websites, but websites earn 27% more value per joule.
- The most powerful model isn't always the best use of energy — task-model matching should weigh efficiency.
- The team's net-value-per-joule (NVJ) metric is already informing where and how they deploy self-improving AI in the real world.
A quantifiable path to answering "how much economic value does AI's energy consumption actually buy?"
More from AGI Musings
- Scott Alexander clashes with critic over whether AI labs may race without public consent — wfithian · 2026-11-07
- A 2x capabilities speedup, 3x algorithm progress: Tom Davidson says this is what 2027 IQ explosion looks like — eli_lifland · 2026-11-07
- AI labs pay up to $2M for corporate work files and emails, data trade may rival compute — pmddomingos · 2026-11-07
- Delvetown: an internet town where humans and AI agents live as neighbors hits 1,000+ residents — Miles_Brundage · 2026-11-07
- ICLR reviews flagged 91-93% AI-generated, sparking peer review integrity debate — 3scorciav · 2026-11-07
- Critic: Anthropic backs recursive self-improvement at full speed while only talking about safety — peterwildeford · 2026-11-07