FaceDeer

joined 1 year ago
[–] [email protected] -2 points 9 months ago (2 children)

Why are you so completely confident that your take on OP's statement is correct and my take on OP's statement is wrong? You're not Mr_Blott.

[–] [email protected] -1 points 9 months ago (1 children)

This is the most frustrating thing, so many people are arguing against their own interests with their efforts to "lock down" their content to prevent AIs from training on it. In this very thread I've been accused of being pro-giant-company when I'm quite the opposite. The harder we make it to train AI, the stronger the advantage that the existing giant companies have in this field.

[–] [email protected] 0 points 9 months ago

Thanks. Sometimes a randomly-chosen name from 13 years ago just takes on a life of its own over time. :)

[–] [email protected] 0 points 9 months ago

AI trainers do a lot of work filtering and reformatting the training data. Often that's the most expensive part. There's a lot of synthetic data used these days too, reprocessed by other AIs.

[–] [email protected] -2 points 9 months ago (10 children)

You can't put conditions on it retroactively. You already published.

[–] [email protected] 11 points 9 months ago (1 children)

Negative examples are often just as useful for training an AI as positive ones. And it all depends on what you want to use the AI for. A moderator bot, for example, needs familiarity with the whole range of user responses it might see.

view more: ‹ prev next ›