Cohere Labs' ATE dataset finds only 2.6% of agentic tools match real work tasks
Cohere · youtube · 2026-10-06
Cohere Labs researchers Campbell Lund and Zanele Munyikwa present the Agentic Task Ecosystem (ATE) dataset, a supply-side measure of automation that maps agentic tools to occupational tasks.
- Headline finding: only 2.6% of agentic tools actually match existing work tasks
- The talk covers how the dataset was built, cross-industry matching patterns, and implications for the future of work
- Companion blog post: cohere.com/blog/automations-early-footprint
More from AGI Musings
- Peter Norvig on 40 years of AI: why experts get the future wrong — TheTuringPost · 2026-10-06
- 8000+ mathematicians sign anti-AI open letter; Conjecture pushes back — burny_tech · 2026-10-06
- Mercor CEO on 60 Minutes: AI is 'changing human work', 100k freelancers now train models — briannekimmel · 2026-10-06
- Epoch AI: compute could soon support hundreds of millions to billions of AI agents — Jsevillamol · 2026-10-06
- AI as a 'generalized bottleneck breaker': why it may differ from all prior tech — realcleverscience · 2026-10-06
- Discovery curves update: cyber accelerating fastest, math bending further, algorithms now bending — dfrsrchtwts · 2026-10-06