Hacker News new | past | comments | ask | show | jobs | submit login
Google asking employees to help train Bard by rewriting wrong answers (9to5google.com)
14 points by mikece on Feb 16, 2023 | hide | past | favorite | 3 comments



So, the employees are asked to improve the quality of their training data based on model inference results. To me, this just looks like the scientific method.


At least some, maybe most AI is just people typing shit in, same with the art. I've discovered clip art copy pasted into "Stable Diffusion" art with simple select / replace. As if a middle school aged child had simply done it.


Is there a way for outsiders to beta test Bard? I'd love to test it.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: