Giuliano and Design-l Group;
I confess I don't thoroughly understand the definition of "intelligent
systems." However, I am interested in the ways computers can participate
in a decision-making process beyond offering useful information in an
effective format.
At the moment, I am willing to consider chance operations, the random
combination of formal elements [see my book _The CAD Design Studio_,
McGraw Hill, 1991 for a crude example]. If, in the process of design, I
can give myself a range of choices I wouldn't ordinarily consider, I
might be stimulated into a new train of thought or the limitations of my
initial assumptions might become apparent.
If we are after discoveries, rather than solutions, apparently unrelated
material need to be understood in the same framework. I'm afraid that
the kind of "evaluation function" a computer may apply will, necessarily,
be based on some codification of prior experience (human or
computer(?)). Again, without understanding the potential of intelligent
systems, I would doubt that such an evaluation would be able to detect
the potential implied by the possible conjunction of two imperfect
solutions.
From a pragmatic point of view, maybe there are scales and aspects of
environmental design which one would be willing to give up to be
able to focus on aspects of greater significance (sic)? I would rather
hope we develop a concept and techniques permitting a partnership between
computer and designer in which such Solomonic questions can be avoided.
Surely, there will be teams of person+computer teams permitting more
effective and imaginative participation which could avoid the
generalizations and simplifications which have destroyed our architecture
and, perhaps, to a lesser degree, our interior design.
Ciao,
Steve Jacobs
----------------------------------------------------------------------
Stephen Paul Jacobs | School of Architecture
spjacobs@xxxxxxxxxxxxxxxxxxxxxxx | Tulane University
(fax)1-504-862-8798 | New Orleans, LA 70118
----------------------------------------------------------------------
On Fri, 30 Sep 1994, Giuliano Bossi (TheBox) wrote:
> Steve,
>
> I agree with you when you talk about the relationship between computer (say,
> expert system) and designer. I'm thinking of a model in which human
interaction
> is essential. You write about the generation of alternative combinations. I've
> found some architects that have no trust in the computer capability of such a
> generation: to make alternative generations the computer must have some kind
> of evaluation function. They don't think that this concept is ready to
> formalize and introduce into an expert system.
> By now, I'm thinking about a system that can produce a combination to fit
> some constraints with qualitative reasoning. The human interaction is intended
> to relax or modify constraints, letting the system producing different
> projects.
>
> What do you think about it?
>
> Bye, Giuliano
>
> --
> Piergiuliano Bossi
> e-mail: bossi@xxxxxxxxxxxxxxxxxx
> fidonet: 2:331/308.8
>
I confess I don't thoroughly understand the definition of "intelligent
systems." However, I am interested in the ways computers can participate
in a decision-making process beyond offering useful information in an
effective format.
At the moment, I am willing to consider chance operations, the random
combination of formal elements [see my book _The CAD Design Studio_,
McGraw Hill, 1991 for a crude example]. If, in the process of design, I
can give myself a range of choices I wouldn't ordinarily consider, I
might be stimulated into a new train of thought or the limitations of my
initial assumptions might become apparent.
If we are after discoveries, rather than solutions, apparently unrelated
material need to be understood in the same framework. I'm afraid that
the kind of "evaluation function" a computer may apply will, necessarily,
be based on some codification of prior experience (human or
computer(?)). Again, without understanding the potential of intelligent
systems, I would doubt that such an evaluation would be able to detect
the potential implied by the possible conjunction of two imperfect
solutions.
From a pragmatic point of view, maybe there are scales and aspects of
environmental design which one would be willing to give up to be
able to focus on aspects of greater significance (sic)? I would rather
hope we develop a concept and techniques permitting a partnership between
computer and designer in which such Solomonic questions can be avoided.
Surely, there will be teams of person+computer teams permitting more
effective and imaginative participation which could avoid the
generalizations and simplifications which have destroyed our architecture
and, perhaps, to a lesser degree, our interior design.
Ciao,
Steve Jacobs
----------------------------------------------------------------------
Stephen Paul Jacobs | School of Architecture
spjacobs@xxxxxxxxxxxxxxxxxxxxxxx | Tulane University
(fax)1-504-862-8798 | New Orleans, LA 70118
----------------------------------------------------------------------
On Fri, 30 Sep 1994, Giuliano Bossi (TheBox) wrote:
> Steve,
>
> I agree with you when you talk about the relationship between computer (say,
> expert system) and designer. I'm thinking of a model in which human
interaction
> is essential. You write about the generation of alternative combinations. I've
> found some architects that have no trust in the computer capability of such a
> generation: to make alternative generations the computer must have some kind
> of evaluation function. They don't think that this concept is ready to
> formalize and introduce into an expert system.
> By now, I'm thinking about a system that can produce a combination to fit
> some constraints with qualitative reasoning. The human interaction is intended
> to relax or modify constraints, letting the system producing different
> projects.
>
> What do you think about it?
>
> Bye, Giuliano
>
> --
> Piergiuliano Bossi
> e-mail: bossi@xxxxxxxxxxxxxxxxxx
> fidonet: 2:331/308.8
>