Search papers, labs, and topics across Lattice.
Princeton AI Lab
3
0
6
People systematically overestimate the efficiency gains from using AI for simple tasks, even when it wastes their time.
Even when explicitly warned about potential deception, LLMs can still be persuaded to make incorrect decisions, highlighting a critical gap between task performance and vigilance.
VLMs are nowhere near human-level general intelligence: they score less than 10% of human performance across a diverse set of human-designed games, especially struggling with world-model learning, memory, and planning.
