Post #1174220
2022-12-10 05:12 UTC
Like pretty much everybody tech-adjacent, I've been thinking about the implications of mainstream tools like ChatGPT. Most of what people learn and believe is already mediated through selective algorithms. We know social media can be used to game elections and discourage civil engagement, but AI chat tools could soon affect every single interaction we have on the web. Most of us are bad at assessing the evidence for what we hear or read. (1/6)
Replies (1)
-
@dhobern@scicomm.xyz 2022-12-10 05:13
Web scraping combined with (at least potentially) hidden weightings now produces plausible and seemingly authoritative outputs. Concerns are already being expressed that our trusted knowledge sources risk ingesting "evidence" generated by such AI processes. Trusted knowledge sources are the only basis for training "good" AI tools, so this circularity is alarming. It means that it is now even more important for us to preserve and protect major corpora of data and foundational knowledge. (2/6)