[Haskell-cafe] GPT & Haskell

[Aloïs Cochard]

>From a theoretical point of view, this can be observed in
https://arxiv.org/pdf/2303.12712.pdf from page 126 to the end of the paper.

More concretely the scale at which such a system can be used to synthesis
information or "predict" stock market is alarming, those are just some
random examples.


On Sat, 1 Apr 2023 at 14:57, MigMit <migmit at gmail.com> wrote:

> Why is it obvious?
>
> > On 1 Apr 2023, at 14:51, Aloïs Cochard <alois.cochard at gmail.com> wrote:
> >
> > Very interesting, thanks for sharing.
> >
> > This is a very radical viewpoint, I think there is tons of research that
> can be done on alignment and for this reason I don't think we should shut
> it all down.
> >
> > Obviously it's not a good idea to give access to such a system to
> everyone, I hope that at least we all start to understand this.
> >
> > On Sat, 1 Apr 2023 at 13:52, MigMit <migmit at gmail.com> wrote:
> > Or this:
> https://time.com/6266923/ai-eliezer-yudkowsky-open-letter-not-enough/
> >
> > (to which somebody said "I work in a datacenter. So far no AI threatened
> violence on me; Yudkowsky just did")
> >
> > > On 1 Apr 2023, at 13:36, Aloïs Cochard <alois.cochard at gmail.com>
> wrote:
> > >
> > > I don't know, personally I could not stop laughing at the answer from
> version 3.5... it's so freaking stupid!
> > >
> > > But if like Dominik you are scared after reading that part you might
> be interested to look into:
> > > https://futureoflife.org/open-letter/pause-giant-ai-experiments/
> > >
> > > On Sat, 1 Apr 2023 at 11:44, Dominik Schrempf <
> dominik.schrempf at gmail.com> wrote:
> > > Well, it seems like GPT4 gets most of the questions right. This was
> more of a scary read than a funny one to me!
> > >
> > > Dominik
> > >
> > > On April 1, 2023 8:18:08 AM GMT+02:00, "Aloïs Cochard" <
> alois.cochard at gmail.com> wrote: How can this be useful when you have to
> anyway review everything is doing as he might to just randomly insert a bug
> or a security flaw???
> > > I prefer to read poems by my human friends.
> > >
> > > I highly recommend starting reading this paper at page 128 instead of
> wasting your time on that prompt:
> > > https://arxiv.org/pdf/2303.12712.pdf
> > >
> > > Be ready for a good laugh
> > >
> > > Sincerely yours
> > >
> > >
> > > On Sat, 1 Apr 2023 at 05:36, William Yager <will.yager at gmail.com>
> wrote:
> > > On Fri, Mar 31, 2023 at 7:27 PM Mig Mit <migmit at gmail.com> wrote:
> > > How is it going to benefit anyone though? English, like all natural
> languages, is terrible for describing algorithms.
> > >
> > >
> > > The other day, I asked ChatGPT to solve a boring but non-trivial
> programming problem I had:
> > >
> > > > I have a png file which contains a bunch of 5x7 bitmap font
> characters. Write a python program which reads in the png, and then for
> each character in the png, extracts the 35 bits of each character and
> assembles them into an integer.  The final output should be a list of
> integers, each integer containing a binary representation of each character
> in the font.
> > >
> > > It did this basically perfectly.  It wasn't precisely matched to my
> format, but only because I was ambiguous in the problem statement. I gave
> it a little more context about the image format and it did it exactly right
> on the second try.
> > >
> > > I also asked it to write an inverse function of this, which it did
> perfectly in Python, and did pretty well in Rust, although with a small and
> understandable bug.
> > >
> > > Very useful!
> > > _______________________________________________
> > > Haskell-Cafe mailing list
> > > To (un)subscribe, modify options or view archives go to:
> > > http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe
> > > Only members subscribed via the mailman list are allowed to post.
> > >
> > >
> > >
> > > --
> > > Λ\oïs
> > > http://twitter.com/aloiscochard
> > > http://github.com/aloiscochard
> >
> >
> >
> > --
> > Λ\oïs
> > http://twitter.com/aloiscochard
> > http://github.com/aloiscochard
>
>

-- 
*Λ\oïs*
http://twitter.com/aloiscochard
http://github.com/aloiscochard
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/haskell-cafe/attachments/20230401/673a9d28/attachment.html>