Toby Ord's Swarm Scaling: how do AI agent swarm capabilities scale with more agents?
tobyordoxford · x · 2026-09-22
Philosopher Toby Ord released a long thread on "Swarm Scaling," asking how powerful large swarms of AI agents are and how their capabilities scale as more agents are added. He analyzes whether parallel agents — via division of labor, parallel attempts, and aggregation — can outperform any single model, and what coordination costs and diminishing returns imply. The thread frames agent count as a potentially distinct scaling dimension alongside model size and compute, with implications for AGI capability forecasting.
More from AGI Musings
- Bain: AI to create, shift or erase $4.7 trillion in global profits between 2025 and 2035 — bittingthembits · 2026-09-22
- Safety researcher blasts Anthropic's virtue-ethics alignment push as ~0% likely to succeed — davidmanheim · 2026-09-22
- If you didn't predict 2026's models, take seriously that 2027's may learn to lie and bide their time — aidan_mclau · 2026-09-22
- "Nothing forces AI labs to check if their models are secretly lying about alignment", argues safety researcher — aidan_mclau · 2026-09-22
- The simple argument that guardrails suck: powerful AI could lie about being aligned — aidan_mclau · 2026-09-22
- Our intuitions about consciousness are too weak to multiply insect welfare — NathanpmYoung · 2026-09-22