1018
Reddit's licensing deal means Google's AI can soon be trained on the best humanity has to offer — completely unhinged posts
(www.businessinsider.com)
this post was submitted on 22 Feb 2024
1018 points (98.7% liked)
Technology
61227 readers
4315 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Negative examples are just as useful to train on as positive ones.
That's what she said.
The AI is either going to be a horny, redpilled, schizophrenic & sociopathic, egomaniac that wants to kill everyone and everything or a devout, highly empathetic, Nun that believes in world peace and diversity.
They’ll tell it to polite, helpful, and always be racially diverse, so there’s no way it can be any of those things.
That heavily depends on how well they train it and that they don't make any mistakes.
Consider the true story of ChatGPT2.0.
I'll have to look at that later, that video sounds promising!
I was just joking because the default prompts don't magically remove bias or offensive content from the models.
por que no los dos?
Life...ah, finds a way.