[eu-gene] Generative or not? cubo23

[Jim Andrews]

> > > True, we haven't yet found a reason why computers can't think or
> > > program themselves.  But I think the bigger problem is that there's
> > > also nothing theoretical that allows machines to program themselves
> > > either.
> >
> > Your second sentence is false.
> > Antoinne gave an example of a programming language that supports
> > self-modifying code.
>
>
> I already made a distinction between self-modifying code and
> programming.  When I said programming, I mean writing something new,
> with intent.

Intent?

> There's nothing at all special about self-modifying
> code.  I've written self-modifying code.  It is no closer to
> artificial intelligence than static code.

Your last sentence is true. But many folks, like the software art critic I
mentioned, don't even realize that self-modifying code is possible. They
think of computers as glorified non-programmable machines.

> But no computer has ever
> actually programmed itself, and noone has ever contructed any theory
> that proves this will ever be possible, or provides any mechanism or
> framework for it.

What do you mean by "programmed itself"? Of course nothing can create
itself. But things can usually modify themselves in some ways. Software can
modify itself completely. Radically. From anything writable to anything else
writable.

> It may in fact be possible, but whether it is
> possible is not yet known.

As mentioned above, I'm not sure what you mean by "program itself" apart
from "modify itself". And the latter is definitely possible.

> The fact that it hasn't been proven
> impossible does not mean that it is or will be possible, it could mean
> that the proof of impossibililty will be discovered ten years from
> now, or it could mean that this subject it too complicated, and we'll
> never figure out whether its possible or not.

I am still curious about what you mean by "program itself".

> This is analogous to the earlier discussion about interactivity versus
> reactivity.  When Alex's AI guru friend said no computer has ever been
> truly interactive, he's simply saying that true machine consciousness
> has never existed yet, and he's right.  The software that we currently
> call interactive, are interactive in the same way that a rock hitting
> water is interactive.  The rock and water react to each other in
> physical ways and change each other's movements, but is there meaning
> in the interaction, are they having a conversation?

I think it's generally accepted that there is no software yet that is
anywhere close to human in consciousness.

But I have experienced interactive work in which there is more meaningful
communication going on than happens in many conversations. And that
communication is sort of between the programmer and the player, but the
programmer isn't anywhere to be found, of course. We are all familiar with
dramas in which someone is given clues and persues a mystery or several, and
they do not have the means of entrance into B before they figure out A. And,
along the way, they figure out 'the meaning of it all'.

I would call this sort of inferential activity interactive if it goes deep
enough, and it sometimes does.

The life of a work of art is the life of art. Not machine consciousness. But
lively nonetheless.

ja
http://vispo.com