Definition
Plain language
Teaching an AI not a specific skill but the skill of getting better at new tasks quickly.
As stated in the literature
Learning-to-learn: optimizing a model so it adapts rapidly to novel tasks or environments rather than mastering any single one; the framing behind training agents to take useful notes that help their future selves.
Also called: learning-to-learn, meta-reinforcement learning
Why it matters: It aims at adaptability itself, so a system can handle unfamiliar tasks quickly rather than needing full retraining for each one.
For example, instead of training a model to play one game, you train it so that it picks up any new game faster than before.
Heard on the show
“The other meta-learning-for-LLMs systems get discussed and differentiated in prose, but nobody's run head-to-head on the same tasks.”Episode 160 — Training an AI to Take Its Own Notes, So Its Future Self Works Better