Today in "AI Ethics." A YouTuber trained a language model on millions of 4chan posts and released it publicly. It has already been downloaded 1.5k times. One user,@KathrynECramer, tested it a few hrs ago by prompting it with a "benign tweet" from her feed. Its output: the N-word.
— Arthur Holland Michel (@WriteArthur) June 7, 2022
Today in "AI Ethics." A YouTuber trained a language model on millions of 4chan posts and released it publicly. It has already been downloaded 1.5k times. One user,@KathrynECramer, tested it a few hrs ago by prompting it with a "benign tweet" from her feed. Its output: the G*mer Word
https://twitter.com/WriteArthur/status/1534104189382676480
- 60
- 103
Jump in the discussion.
No email address required.
@BrokeBackBuck How hard is it really to train a model? I mean downloading the data is easy, and then you just put it into the same program right? Making the program is the hard part
Jump in the discussion.
No email address required.
Anyone who can use a computer can train a basic model with a little effort. Like how anybody can learn Excel. It might not be very good, if you don't know what you're doing, but still.
Most Reinforcement Learning researchers can't code for shit, because they can just rely on stuff other people have written. That's the main reason I stay away from RL.
Jump in the discussion.
No email address required.
so this youtuber basically did nothing and is full of shit? like clock boy
Jump in the discussion.
No email address required.
Just because it's easy doesn't mean it's wrong. And they did probably have to do a lot of work to get the data in the right format.
tbh if it started screaming about BIPOCs that's probably a sign it's actually working
Jump in the discussion.
No email address required.
More options
Context
More options
Context
More options
Context
More options
Context