[seqfan] Thread on AI chatbot

[Olivier Gerard]

Dear Members,

Here we are going too far from our subject, which is integer sequences.

These systems have limitations, biases and flaws, which are already
quite well-known and documented.
They have applications and qualities as well.
Some people on this list have a long experience building them or testing
them.

But this is not the purpose of this list to discuss this.

You may have noticed that the first messages on this thread
were investigating the understanding of statements about sequences by a
particular system.
This was relevant. Stay focused, please.

Thread closed now.

Olivier GERARD
Administrator

On Sun, Jan 29, 2023 at 2:36 PM Antti Karttunen <antti.karttunen at gmail.com>
wrote:

> There is a fatal flaw in this kind of AI: apparently it is impossible
> for it to admit, "I don't really know."
>
> Still, I have now heard from my colleagues, how they have switched
> from using Google search to ChatGPT, "as it knows everything so much
> better".
> Yes... what could go wrong?
>
> Antti
>
>
> On 1/28/23, Fred Lunnon <fred.lunnon at gmail.com> wrote:
> > My immediate reaction too!     WFL
> >
> >
> >
> > On Sat, Jan 28, 2023 at 9:08 PM Marc LeBrun <mlb at well.com> wrote:
> >
> >> >=Allan Wechsle
> >> > You can tell it doesn't really understand what it's saying,
> >> > but it still provides interesting ideas for investigation.
> >>
> >> A colleague offhandedly described these chatbots as "like a bad grad
> >> student, who will lie to make you happy"
> >>
> >>
> >> --
> >> Seqfan Mailing list - http://list.seqfan.eu/
> >>
> >
> > --
> > Seqfan Mailing list - http://list.seqfan.eu/
> >
>
> --
> Seqfan Mailing list - http://list.seqfan.eu/
>