In this episode of AI Data Chats, we are joined by Justin Zhao, independent researcher (ex-Google, ex-Predibase). Justin has worked with some the leading AI research teams and his independent research over the past year has included a paper on the Language Model Council. In our chat, Justin shares his thoughts on why LLM as a Judge may not be good enough. By leveraging a jury of models, you can mitigate biases, hallucinations or bad decisions of a single model. Justin also shares some interesting findings on the voting preferences of different model families.

Speakers

Jennifer Ding

Solutions Engineer @ Encord

Justin Zhao

Independent researcher (ex-Google, ex-Predibase)