3 ms·

Another person who doesn’t realise AI language models are just making shit up. Google results are quite often full of wrong information, but at least it has mec

by Veen 4y ago

Another person who doesn’t realise AI language models are just making shit up. Google results are quite often full of wrong information, but at least it has mechanism for surfacing better content: inbound links, domain authority, and other signals. It doesn’t guarantee correctness, but it’s better than the pseudo-authoritative fiction GPT-3 and friends come up with.

cwkoss 4y ago
Most of the time, humans just make shit up too. I just made up the contents of this comment.
- ep103 4y ago
  Thankfully, your comments aren't being used as a definitive source of truth by default :)
  - cwkoss 4y ago
    Unfortunately, this applies to the vast majority of appeals to human expertise, though.
    - ep103 4y ago
      Yup, which is why what we socially define as reputable (read, sacred or protected) sources and contexts are so important : )