‹ BackHN Continuity

Thread

Introducing System One Models and Jev

1989 points · 520 comments · albelfio

  1. wg0 · · focus · HN ↗
    Can I put it as Air Traffic Controller? With similar error rates as humans?

    That would be the litmus test.

    "Does not hallucinate" is not the same as "is never wrong".

    So the ATC test could be the benchmark.

    1. copperx · · focus · HN ↗
      Not hallucinating is easy when you don't produce strings.
      1. vintermann · · focus · HN ↗
        Hallucinating as we use the word really only applies to generative AI. Non generative AIs can't hallucinate, they can just be wrong.
Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.