Search papers, labs, and topics across Lattice.
1
0
3
1
Even state-of-the-art vision-language models frequently lie and hallucinate when playing social deduction games, raising serious questions about their reliability in real-world applications requiring grounded reasoning.
