"Ethics" is not the only problem with generative A.I. -- *epistemology* is broken too.
My latest for Public Books: in the face of #google telling us to eat rocks, we bought an encyclopedia.
https://www.publicbooks.org/the-encyclopedia-project-or-how-to-know-in-the-age-of-ai/
Companion blog post here:
"Yet knowledge is not a market commodity. Moreover, “justified true belief” does not result from an optimization function. Knowledge may be refined through questioning or falsification, but it does not improve from competition with purposeful nonknowledge. If anything, in the face of nonknowledge, knowledge loses."
FYI @benjedwards @justinhendrix @JustinPotBlog @emilymbender @pluralistic @aram @randomwalker @histoftech @eszter @natematias @ntnsndr @Wolven
@cyberlyra @benjedwards @justinhendrix @JustinPotBlog @emilymbender @pluralistic @randomwalker @histoftech @eszter @natematias @ntnsndr @Wolven I wish everyone could internalize this concept
@cyberlyra This looks awesome. I'm glad to see such perspectives.
@cyberlyra I wonder if one of the biggest problems with the hype around LLMs as "knowledge" (e.g. Perplexity's website) is their divorce from the social, or rather, their embeddedness in potentially dystopian alternatives. (To riff off Ludwik Fleck's observation that "Cognition is the most socially-conditioned activity of man, and knowledge is the paramount social creation [Gebilde].")
@cyberlyra It’s a pity that Microsoft is so all-in on AI. A trustworthy multimedia encyclopaedia like Encarta would be immensely useful right now.