you are viewing a single comment's thread
view the rest of the comments
[–] 1 point 11 months ago (3 children)

Our thought processes contain mechanisms that are not well understood, to the point where LLMs (as a side effect of human programmers’ inability, not to their discredit) are unable to effectively mimic them. An example of this would be ChatGPT encouraging self-harm where a human assistant intelligence would not.

  • source
  • parent
  • hideshow 3 child comments