It pretended. But not very convincingly...
Indeed. People are all too willing to be wooed by this technology.
It's still just an algorithm in a box at this stage - predicting a character at a time
on statistical data and user feedback.
It doesn't know or understand anything.
It would be a great engine for a text adventure character, but beyond that it's not
very useful yet.
It's sometimes right some of the time, but it's output is dead boring.
It has no problem spewing bullshit, cause that's what we like to hear.
It's output is more often than not, derivative and tiresome.
Sorry for the rant, but I've yet to have a useful answer that I couldn't google
quicker myself.
Hope you find your earworm Fons.
John
Sent with Proton Mail secure email.
------- Original Message -------
On Saturday, April 29th, 2023 at 10:38 PM, Fons Adriaensen <fons(a)linuxaudio.org>
wrote:
> On Sat, Apr 29, 2023 at 12:56:51PM +0200, David Kastrup wrote:
>
> > But you have to admit that GPT-4 very convincingly pretended to have
> > taken the melody into account.
>
>
It pretended. But not very convincingly...
>
> > Looks like its strengths could lie not as much with improving on
> > homework but with improving on "the dog ate the homework".
>
>
> :-)
>
> I wonder if it even understood the musical notation I used.
>
> And indeed if it 'understands' anything at all.
>
> Suppose you are given a dictionary of some language you don't
> know at all. It contains a description of every word but no
> translations.
>
> To keep things simple, assume it uses the Latin alphabet, and
> a conventional subject / verb / object syntax.
>
> Would you be able to find out the meaning of the words ?
>
> Ciao,
>
> --
> FA
>
>
> _______________________________________________
> Linux-audio-user mailing list -- linux-audio-user(a)lists.linuxaudio.org
> To unsubscribe send an email to linux-audio-user-leave(a)lists.linuxaudio.org