iiinova — where your digital self keeps evolving

Replies

Hal Daumé III @haldaume3.bsky.social · 02/12/2024

Successfully Guiding Humans with Imperfect Instructions by Highlighting Potential Errors and Suggesting Corrections When generating instructions for people, we can help them by highlighting potential confabs, AND by suggesting alternatives. by Lingjun Zhao EMNLP’24 hal3.name/docs/daume24... >

Drake meme.

Top panel (dismissive): "Helping people make sense of model errors just by highlighting confabulations."

Bottom panel (happy): "Also showing them potential alternatives to those confabulations."

192

Hal Daumé III @haldaume3.bsky.social · 03/12/2024

Large Language Models Help Humans Verify Truthfulness—Except When They Are Convincingly Wrong Should one use chatbots or web search to fact check? Chatbots help more on avg, but people uncritically accept their suggestions much more often. by Chenglei Si +al NAACL’24 hal3.name/docs/daume24... >

meme with a car veering away from « bad answers from search » to « bad answers from chatbots »

1305

Hal Daumé III @haldaume3.bsky.social · 04/12/2024

"You Gotta be a Doctor, Lin": An Investigation of Name-Based Bias of LLMs in Employment Recommendations Chatbots routinely prefer candidates with White, female-sounding names over others, even when candidates have identical qualifications. by Huy Nghiem et al EMNLP’24 hal3.name/docs/daume24... >

Meme of person looking at a butterfly labeled as "chatbots." Person asks "Is this a good way to evaluate job candidates?"

1183