3 ms·
Google relies on human employees to improve Bard chatbot responses
- isaacfrond 4y agoSounds a lot like the Reinforcement Learning from Human Feedback (RLHF) that ChatGPT was trained on. Seems like such a training phase is needed to avoid the Sydney/Bing search debacle.