This research investigates how language models balance autonomous action against human escalation, revealing critical flaws in their self-assessment of corre...
Level: advanced
By Matthew DosSantos DiSorbo, Harang Ju
Category: research