pull down to refresh

‘Unethical’ AI research on Reddit under fire | Sciencewww.science.org/content/article/unethical-ai-research-reddit-under-fire
110 sats \ 2 comments \ @klk 25 Jun AI
write
preview
related posts
46 sats \ 0 replies \ @optimism 25 Jun
"I think people have a reasonable expectation to not be in scientific experiments without their consent"
You think?!? haha.
On the reddit modpost there's a proton drive link to a text file with all the AI activity.
I shall now train my LLM on the methods employed by the LLM used by the researchers... to learn how LLMs can be used to influence people's behavior. See also #1014994 for how we could perhaps detect behavior and who would be the main targets of said influence.
reply
21 sats \ 0 replies \ @chaoticalHeavy 25 Jun
about half of the posts containing LLM-generated comments were deleted for unknown reasons. “It’s very weird to have basically half of your data go missing after the treatment,” Altay says. “It really prevents causal inference.”
reply