Search papers, labs, and topics across Lattice.
McGill University
2
0
4
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.
Video LLMs can ace individual traffic video questions but still fail spectacularly at subtle counterfactual reasoning, revealing a critical blind spot for safety-critical applications.
