Comments by "Guinness" (@GuinessOriginal) on "Michio Kaku on Quantum Computing and AI Chat-bots" video.

  1. AI LLMs like ChatGPT can discern what’s accurate, you just have to set it up and prompt it the right way.
    1
  2. Auto generator can code a working version of snake from scratch by just giving it the requirements - and you can tell it to get them from the web.
    1
  3.  @UnchainedEruption because a) or using an LLM trained survivor on language first, not a dedicated code AI model trained specifically to do code; b) Chat GPT was trained to code on stack overflow code snippets, which are often buggy. So it’s been trained to write buggy code.
    1
  4.  @RorschachRev what do you do?
    1
  5.  @MattSociable I’d check those sources sporadically. Unless you’ve set up your model parameters to be conservative and deterministic, and have promoted it to be factual and Al evaluate its answers for accuracy, it will still either make up a source or give a real source but lie about the information contained in that source. I caught Bing doing it yesterday, the cheeky rogue.
    1
  6.  @Gcssdvnkloiutesc great point. AI has the danger of dumbing us down even further. Although let’s be honest, most of America is already dumbed down beyond belief, so it might actually raise the intellect of most of it
    1
  7. He also didn’t mention that you can Prompt it to fact check and get it to evaluate the answers it gives for accuracy.
    1
  8.  @AbadonBIack Joe regularly gets his waters muddies 😂 and he’s explaining it for Joe’s audience. Joe is just a conduit for them. Think of Watson in the Sherlock Holmes novels.
    1
  9.  @sobhhi oh, and you can prompt it to fact check, and evaluate it’s answers for accuracy
    1
  10.  @richrdroma only if you set the parameters of the model for it to be creative. You can also set it up to be conservative and deterministic.
    1
  11.  @AbadonBIack I’m afraid to inform you that you are incorrect. ChatGPT and LLMs can indeed be creative and original.
    1
  12.  @fireteamomega2343 it definitely doesn’t, at all
    1
  13.  @Sekhmet6697 you’re asking an AI trained on language to do maths. There are maths AI for that, which you can hook Up to Chat gpt in the background and it can just use them instead seamlessly. In any case, with the right prompt you bag get it to give you the right answer. It’s just a really inefficient way of doing maths.
    1
  14.  @Sekhmet6697 you’re asking an AI trained on language to do maths. There are maths AI for that, which you can hook Up to Chat gpt in the background and it can just use them instead seamlessly. In any case, with the right prompt you can get it to give you the right answer. It’s just a really inefficient way of doing maths.
    1
  15.  @AbadonBIack  I understand exactly how they work. Not only do you have a limited notion of how they work, you also don’t really understand how the human brain works. It’s fine; you want to believe they can’t be creative. Ignore the facts, just regurgitate some soundbite you heard on the MSM. It’s easier to fool people than convince them they’ve been fooled, especially when they are so keen to fool themselves that they’re the sharpest tool in the box.
    1
  16.  @AbadonBIack  no. You said he doesn’t want to muddy the water for Joe. I explained that he knows he’s really talking to the audience and in fact he doesn’t want to muddy the waters for people like you.
    1
  17.  @AbadonBIack  I don’t think so, but don’t worry about it. You can believe whatever you want, including that you’ve got the biggest brain in the YouTube comment section. Hey, as long it makes you happy bub.
    1
  18. You want to use code interpreter for that, or there’s a tool out there called auto generator which is also very good. Both are based on GPT 4.
    1
  19.  @djbryson gpt4 is the upgrade to chatGPT and is connected to the internet.
    1