LLMs (including Deep Research) assume your question is coherent or a good one.

· Bits and Bobs 3/10/25
  • LLMs (including Deep Research) assume your question is coherent or a good one.[vc]
    • It's very easy to accidentally trick yourself with some superficially good output on a fundamentally flawed question.
    • Model: "Turns out you were right all along!"
    • Human: "Just as I expected, thank you!"

More on this topic

From other episodes