Glossary term
Glossary term
Foundations
An LLM's ability to generate responses for prompts that it was not explicitly trained on.
Created for this library
An LLM evaluation team tracks emergent behaviors like multi-step arithmetic as models scale to plan which capabilities to publicize in product releases.
A research lab studies emergent behavior in agent systems by varying compute and scaffolding to identify what enables new capabilities.
A frontier-AI safety team red-teams new model checkpoints for emergent behaviors that were not present in smaller predecessors.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License