UK AISI chart suggests AI time horizons scale as a power law with training tokens
tobyordoxford · x · 2026-10-09
Drawing on a chart released by UK AISI, Toby Ord notes that each model's 80% human-equivalent time horizon follows a roughly straight line against training tokens, implying a power-law relationship: time horizon ∝ tokens^γ.
This means AI models' time horizons scale sub-linearly with tokens, offering a quantitative comparison between how model capability scales and how human time horizons grow.
Related event: UK AISI Data Shows Model Time Horizons Scale Sublinearly with Tokens(4 posts)→
More from Research
- LabCompass preprint uses generative inverse design to steer human hematopoietic cell fate in vitro — anshulkundaje · 2026-10-09
- AgentGarten: Code Worlds with a Real-Time Neural Renderer for Evolving Agents — Scobleizer · 2026-10-09
- Samsung open-sources LittleBit: 13B LLM squeezed under 1GB at 0.1 bits per weight — CurieuxExplorer · 2026-10-09
- AgentTime benchmark finds AI agents can't manage runtime and sometimes sleep to pad hours — mikeflache · 2026-10-09
- Mike Frank: even O(n log n) multiplication isn't practical, new results won't be either — MikePFrank · 2026-10-09
- Renmin University open-sources EvoOntology, a self-evolving ontology layer for data agents — blaizedsouza · 2026-10-09