Robot Dreams

Brian zhil at online.no
Tue Jul 17 13:21:31 EST 2001


"Bloxy's" <Bloxy's at hotmail.com> skrev i melding
news:f52chf9arat.wpkvctm at yqllkg.edu...
> Internal thought process is a mechanism for the biological
> intelligence to EXPERIENCE the physical domain and learn
> the nature of its own entity and life force as such.
>
> It is simply inapplicable to a machine, no matter what kind
> of software you are going to put in.
> No matter how hard you model your most intricate mechanisms,
> it can not possibly function as intelligence.
>
> You simply removed the CORE of it.
> All you have is a plastic lable on a bottle of coca cola.

Agreed.

> > even
> >if external thought processes are not present at a given time. However
> >due to the nature of neurons, any neuron that is not used will deevolve
> >from its network. That includes neurons and neuron groups evolved to
> >serve connections to sensory inputs.

Will Artificial Neural Networks devolve ?
Or wouldn't it oscillate around the latest set-point ?
No, I don't think it will devolve, but it certainly won't evolve as it isn't
dynamic as the biological intelligence.

> >If we want the ANN system to be adaptable, we cannot allow it to
> >perform endless thought processes if no sensory inputs are arriving
> >from the outside world.
>
> This is merely the issues of efficiency
> and not the issue of principle.
>
> You are merely optimizing the performance
> or conserving energy, whatever pleases you more.
>
> It does not change ANYTHING at the core of the issues.

Agreed Bloxy's.
Your post was highly informative and beyound the scope of
primitive intelligence.

Brian





More information about the Neur-sci mailing list