Search papers, labs, and topics across Lattice.
3
0
6
1
An AI audit of popular chatbots using real commercial-advice queries shows that neither isolated responses nor API observations can be assumed to represent the commercial advice consumers encounter, and independent audits of AI-mediated commercial advice should account for repeated responses, consumer-facing conditions, and the source layer being observed.
Dutch language models exhibit alarming biases, with some favoring stereotypical representations of transgender identities up to 97% of the time.
Train robust speech translation models without generating adversarial speech data: transfer adversarial robustness from text using Cross-Modal Robustness Transfer (CMRT) and gain +3 BLEU.
