Definition
Plain language
When the situations a system meets in real use drift away from the examples it learned on, so it starts making mistakes.
As stated in the literature
A mismatch between the training distribution and the deployment or visited-state distribution; the core reason behavior cloning compounds error, since flawless experts never demonstrate recovery from off-distribution states.
Why it matters: It matters because systems that look flawless in training can fail badly in the wild once real conditions drift from what they saw.
For example, a self-driving model trained only on smooth highway footage may flounder the first time it hits a snowy back road.
Heard on the show
“In absolute terms, it's about eight points of accuracy on the split that tests distribution shift, and about eleven points on the split that tests recurring workloads — and notice which gap is bigger.”Episode 168 — When Turning Experience Into Code Makes Your AI Agent Dumber