Large Language Models’ Emergent Abilities Are a Mirage

https://media.wired.com/photos/65fcb020abfbcf31f180cf6d/master/pass/AbilitiesJumpMirage-byKristinaArmitage-Lede-scaled.jpg

A new study suggests that sudden jumps in LLMs’ abilities are neither surprising nor unpredictable, but are actually the consequence of how we measure ability in AI.

https://www.wired.com/story/how-quickly-do-large-language-models-learn-unexpected-skills/