This one is also interesting
https://www.emergentmind.com/
My current favorite:
https://www.emergentmind.com/posts/write-me-a-paragraph-as-if-a-baby-accidentally-wrote-it
Edit: This one is great too:
https://www.emergentmind.com/posts/give-instructions-on-how-to-google-something-in-the
@Kingscrusher-YouTube said in #28:
> Any tool has strong and weak points - for me, I think the areas analogous to essay writing will be where it shines. That is also where it is being often abused right now causing concerns in Academia of course.
To be honest... if you're no good at writing essays, this 'tool' sure has hell isn't going to compensate for it. It's just too obviously not capable of anything remotely like that. The only kind of audience of such essays is the kind that doesn't actually try to read and understand the text. It produces 'sophisticated gibberish', not texts that actually make sense.
I don't think of ChatGPT as a tool (as far as I can tell it's perfectly useless for any application) but as representative of the current state of the art when it comes to AI systems trying to have a conversation.
And as said: if that is the state of the art, it's very, very disappointing.
@Molurus said in #32:
> To be honest... if you're no good at writing essays, this 'tool' sure has hell isn't going to compensate for it. It's just too obviously not capable of anything remotely like that. The only kind of audience of such essays is the kind that doesn't actually try to read and understand the text. It produces 'sophisticated gibberish', not texts that actually make sense.
>
> I don't think of ChatGPT as a tool (as far as I can tell it's perfectly useless for any application) but as representative of the current state of the art when it comes to AI systems trying to have a conversation.
>
> And as said: if that is the state of the art, it's very, very disappointing.
Are you aware of developments generally like Copilot - a lot of code is being written with it.
Also for example, even Mosh thinks it can greatly produce programming productivity - "Codex" for programming languages is another aspect of it. See video here:
https://www.youtube.com/watch?v=sTeoEFzVNSc
64 is 8x8. guess what, a byte is 8 bit. a word is 8 byte (i get confused, it might be french for byte, although there is octet).
anyway byte^2 = 64 bits... The end is nigh!
@NHL_20 said in #9:
>
@NHL_20 said in #9:
> TLDR:
>
> ChatGPT reflects quite well the written down human knowledge. Including the errors in that knowledge.
Well, it introduces lots of new errors on top.
Hello @Kingscrusher,
thank you for your interesting article, I really enjoyed it - but maybe it's a bit too long.
Here in the previous 35 posts I couldn't see the following error being pointed out.
You write in your blog:
"Was Kasparov lucky to not lose against Karpov in their first match?
It's not accurate to say that Garry Kasparov was lucky to not lose against Anatoly Karpov in their first World Chess Championship match in 1984-85. The match was a highly competitive and intense battle between two of the strongest chess players of the time, and both players showed great skill and determination throughout the match.
In the end, the match was controversially stopped after 48 games, with the score tied at 5 wins each and 33 draws."
Please correct that the result of this canceled World Chess Championship is:
Karpov 5, Kasparov 3 and Draw 40.
Thanks.
@NHL_20 said in #8:
> I stopped reading at "Who are the most important chess hypermodernists?".
>
> Admittedly, ChatGPT is able to build correct sentences without grammatical errors, which even make very much sense most of the time.
>
> On a closer look, one can see, that 50% of the content is well known knowledge which a Google search will reveal in the first few results. The other 50% is self repeating bla bla. For example a lot of sentences say, in different words: "I am not sure, look for yourself, decide on your own".
>
> ChatGPT does what it is supposed to do: Reflect the available knowledge in the world. It doesnt invent anything new. At least not in these answers.
>
> That is not a critic. On just has to be aware what ChatGPT is: A next generation search engine, which can give impressive answers, but also can contain factual errors. One should always double check instead of blindly relying on it. There is a reason that ChatGPT answers are banned on for example Stack Overflow.
You are absolutely right in your last conclusion. And even the name "neural network" is just as much of a hoax as the loud "AI". Neural networks do not contain any neurons, or any resemblance thereto. Moreover, it can hardly be called a network - there are cascading calculations, not network ones. They tried to make exactly a network (i.e. that connections could be "backwards" or at least horizontal within one layer), but there is such complexity that the task becomes unsolvable. Do I believe in Roko's basilisk? Good question, but I won't answer it.
@Tenakel said in #36:
> (...)
> Please correct that the result of this canceled World Chess Championship is:
> Karpov 5, Kasparov 3 and Draw 40.
> (...)
Good catch, that is another error. But it would be not ok to edit these answers. Because the purpose is to show how the AI performs.
@Kingscrusher-YouTube #12
@CrazyMaharajah #37
It hurts my eyes when you fullquote this way. Please cut out the unnecessary parts.
@Tenakel said in #36:
> Hello @Kingscrusher,
>
> thank you for your interesting article, I really enjoyed it - but maybe it's a bit too long.
>
> Here in the previous 35 posts I couldn't see the following error being pointed out.
> You write in your blog:
>
> "Was Kasparov lucky to not lose against Karpov in their first match?
> It's not accurate to say that Garry Kasparov was lucky to not lose against Anatoly Karpov in their first World Chess Championship match in 1984-85. The match was a highly competitive and intense battle between two of the strongest chess players of the time, and both players showed great skill and determination throughout the match.
> In the end, the match was controversially stopped after 48 games, with the score tied at 5 wins each and 33 draws."
>
> Please correct that the result of this canceled World Chess Championship is:
> Karpov 5, Kasparov 3 and Draw 40.
>
I put a big disclaimer at the start of the blog post now indicating not to trust the accuracy of the answers. I noticed this inaccuracy in the Youtube video at the end - my reactions there included being suspect about the 5-5 - it was 5-3 as you say before the match got abandoned.
Cheers, K