points by simonw 1 month ago

Once you've seen the AI not make mistakes like that a few dozen times you start to trust it not to mess that up in the future.

discreteevent 1 month ago

Is this really a rational strategy for something whose nature is to be right most of the time and then spectacularly wrong a much lesser amount of the time?

  • simonw 1 month ago

    Yes, because the mistakes in code are easy to spot. I wouldn't use this to write me an unreviewed legal brief.

    • discreteevent 1 month ago

      But you said that in this case you don't review the code because you trust the AI having seen it write that kind of code before.

    • twister2920 1 month ago

      you literally just said you don't review this code!