SessionLeadership trackconfirmed
When Will The Benchmaxxing Plague End?
- Day
- Day 2 — Session Day 1 · Tue, Jun 30
- Time
- 2:50pm-3:10pm
- Room
- Track 9 · Room 2016
- Track
- AI Architects: Show my Workflow
Track theme
Data Quality & RL Environments
Data quality for frontier models and reinforcement-learning environments — curation, long-horizon training, and reliability.
Accessible with the Leadership (All-Access) pass and above.
About this session
Model releases are heralded by a flourish of trumpets, a chorus of weeping angels, and often, inflated benchmark claims. Why do benchmarks so often not reflect real-world value? Is it intrinsic to the science of benchmarking, or just the consequence of our current practices? Is LM Arena a cancer on AI?
