Karpathy says only ~5,000 people access frontier AI systems — how would 1,000-agent swarms even work?
dsrtslnd23 · hn · 2026-10-12
- Andrej Karpathy laid out an AI usage stratification: 75% have barely used LLMs, 20% use them casually, 0.2% use frontier models professionally for math and code, and only 0.00006% (5,000 people) have internal access to frontier-grade systems.
- He describes those systems as "swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing mathematics."
- The HN poster asks how such swarms actually work, noting that reliably coordinating even 2-3 agents is already hard — sparking discussion on multi-agent orchestration, communication overhead, and diminishing returns at scale.
More from AGI Musings
- Super-Worms Warning: AI-Powered Self-Propagating Malware and a Fragmenting Internet — JensHonack · 2026-10-23
- Wolfram Opens Rule 30 Prizes to AIs: $30,000 for Three Open Problems — cephaloform · 2026-10-23
- Mathematicians and Biomedical Researchers React to AI in Strikingly Different Ways — manorlaboratory · 2026-10-23
- Stanford and Databricks use AI agents to reproduce medical findings and test analytical robustness — ChengleiSi · 2026-10-23
- ISMS Fellow Avi Goldfarb: AI disruption raises returns to exploring research directions — avicgoldfarb · 2026-10-23
- signull on AI-found PHP exploits: economic liability is the forcing function legacy software needs — signulll · 2026-10-23