Uptodate datasets?

Mmm, it could, but I am not sure that’s where we want to go ?? It’s a bit hacky; and basically what it does, is that it only gets the text of the first result, feed it to ChatGPT so that it makes a simple version of it. You could also copy/paste it. It looks more fun than useful, I think ??

What did you write “Uptodate datasets” as the title?

Thanks.

“What did you write “Uptodate datasets” as the title?”

This biggest issue for my “law bot” attempt is, that it still throw alot of bullshit. Some are just AI limitation, but alot of them seem also to be issue that it only has content from 2021 and earlier yet?

Actually, I think you might need to study a bit more about how AI works; it’s very important to know exactly what it does, how, and also… what it cannot do ?? Have a look here: https://beta.openai.com/docs/guides/completion/prompt-design.

You can’t really “update” the model used by the AI, it’s as it is; you can train it to behave in a different way, with different data, but that involves a lot of work.

The answers given by the AI are hallucinated (https://en.wikipedia.org/wiki/Hallucination_(artificial_intelligence)) and they aren’t really based on facts.

A law bot would be certainly very interested to do; but at this point, I think using the davinci model is the best way, with a good prompt. In the prompt (what you define as “context” in the chatbot of AI Engine), you need to add: If you don’t know the answer, say “I don’t know”. By adding this, you remove a potential issue, but that will not make it perfect.

Thank you Jordy. I have just seen that it often has old data, like if you ask ChatGPT for “What ist the the threshold of…” or so, than it give out old data, despite asking for 2022 or 2023.

But thank you for the links. I will play around with the prompts!