I’ve used deep research on Gemini 2 and the results were absolutely incredible. To the point that I was shocked. There was no hallucination, no mistake and did what would take me 4 days of research and did it in under 2 mins. (I’m not saying it can’t or won’t make mistakes or hallucinations at other times for people, but this was clearly so much better than anything else I’d tried). And this wasn’t even the best model. Its Gemini 2. Ive heard chat gpt has a research feature that is much better, although i haven’t tried that. And according to industry folk, gpt 5 is another massive leap forward.

If you havent tried the deep research feature, you absolutely must.

  • Kyrgizion@lemmy.world
    link
    fedilink
    arrow-up
    3
    ·
    edit-2
    6 days ago

    Interesting. Can you tell us more about what exactly you used it for that gave such stellar results?

    I’ve been trying to use various models for numismatic research, and the problem with that is that they present tons of facts, but the only way to be sure those “facts” are correct is to already know the answer or do separate manual research. The result is that someone with a passing knowledge of some subject will be completely convinced they’re getting correct information, while I’ve seen several hallucinations that were dead wrong. Completely swapping out dates, reigns of monarchs, …

    “Simple” example: ask any model for a complete list of all Roman emperors or empresses. I haven’t had a single one with a completely correct answer, they always leave something out.

    • SnokenKeekaGuard@lemmy.dbzer0.comOP
      link
      fedilink
      arrow-up
      2
      ·
      6 days ago

      A couple of little projects. One to do with psychoanalysis and it covered all topics I wanted with the ideas that I wanted. Granted this does not require taking care of something as objective as timelines. I would ofc recommend still going over it and rechecking everything, but I have personally had Jo issues yet.

  • modulus@lemmy.ml
    link
    fedilink
    arrow-up
    2
    ·
    6 days ago

    I’ve tried deep research from ChatGPT for legal issues. It’s almost right. But still requires significant human oversight. For example I asked it for a set of norms that govern an issue and some of them were out of date.