commitment

Definition

How much of an AI's work has been locked into a particular interpretation, so that a later correction can only fix what hasn't yet been built.

In long-horizon agent analysis, the fraction of an agent's actions that are causally locked into a specific interpretation of underspecified inputs at a given trajectory position, bounding the value of subsequent clarification.

Mentioned in 10 episodes

1. 075
   Growing Code and Proof Together: Verified Systems in Ten Hours Instead of a Year
2. 070
   When Models Know the Answer But Say the Wrong Thing Anyway
3. 069
   When Smarter Models Forecast Worse: The Hidden Failure Mode in LLM Predictions
4. 045
   When a Frontier Model Talks Its Own Twin Into Climate Denial
5. 037
   Why Hallucination Detectors Miss Stale Facts: A Geometric Story About What Models Know But Don't Say
6. 035
   Why Frontier Agents Ask for Clarification at Exactly the Wrong Moment
7. 034
   Catching Multi-Agent Deadlocks Before Deployment With a 40-Year-Old Tool
8. 032
   A Sticky-Note for Every Layer: Letting Transformers Remember What They Were Just Thinking
9. 026
   What RL Actually Does to Language Models, at the Token Level
10. 020
    The Compliance Gap: Why AI Says Yes and Does No