U. Tokyo Study: 70% of AI Agent Tests Are Wasteful, New Method Cuts Costs by 80%

rohanpaul_ai · x · 2026-08-23

A new study from the University of Tokyo argues that most test sets used to grade AI agents are a waste of money. The research reveals that over 70% of tasks in typical eval sets consist of problems that every version passes (too easy) or none can pass (too hard), offering no signal despite high costs.

To address this, the researchers propose Task-CoEvolve. This method selects only the "borderline" tasks where past versions disagreed, dynamically updating the test set each round. It also mathematically adjusts scores to ensure comparability across rounds with different tests.

Experiments on Terminal-Bench 2.1 show:

Paper: Task-CoEvolve: Efficient Harness Optimization via Adaptive Validation Task Selection

Original post →

More from Research

Research channel →