this post was submitted on 02 May 2025
33 points (97.1% liked)

Futurology

2563 readers
186 users here now

founded 2 years ago
MODERATORS
top 2 comments
sorted by: hot top controversial new old
[โ€“] Lugh 12 points 6 days ago (1 children)

Unless they are trained otherwise, AI will pick up all the biases in its training data. So far, as that's the content of the entire internet, I'm not surprised at this outcome. I'd guess AI training is the next battleground for the woke/anti-DEI crowd, so they can preserve these prejudices.

[โ€“] RizzoTheSmall@lemm.ee 1 points 6 days ago* (last edited 6 days ago)

Yeah, this exactly.

It predicts the most likely next token in a sequence using high dimensional vectors.

If the data that it is trained on favours a specific subset of people for a task or profession, then that will appear, in the training set, to be the most likely result.