Tag
This paper investigates mismatches between expected and actual time savings when using LLMs for simple cognitive tasks, revealing a speedup illusion where users underestimate AI-assisted completion times despite no actual speedup.