|
Dear PhD.Design
I’d like to share a link to a paper on generative design I wrote a while back and some more recent thoughts I’ve had about it.
https://www.euppublishing.com/doi/abs/10.3366/dls.2012.0047
The paper uses a Deleuzian-Guattarian framework to understand the creative search/solution space in humans and computers, and I came to Terry's position that algorithms have a vast capacity for generating novelty, but however that this novelty can be both exhausting and self-defeating. I would say humans, despite or probably because of the fundamental property of nature to always change, tend to prefer less novelty than more of it, attributing names and ideas to objects and things in ways that convey stasis, though such stasis is contrary to the physical observation of the world. There really are no two chairs that are exactly the same at a molecular or philosophical level, much less two chairs built in different centuries. Yet the word ‘chair’ groups them together them and attributes the quality of sameness, a ‘molar’ perception. Algorithms don’t have any sensitivity or concern for this cultural construction, beyond any specific parameters that may or may not programmed into them, for example, the need to create a structure that supports the sitting human body in some way.
We are only at the beginning of the information/algorithm age, and don’t have clear idea of how much novelty generative design will produce in the world, especially as algorithms are used increasingly with forms of robotic production that de-privilege the efficiencies of manufacturing identical objects. With generative control over physical matter, we are heading towards a radical reshaping of the planet. Unfortunately this isn’t about good or bad design, however Ken or Klaus want to define it. Rather, it will happen because it can. I tend to think that humans find the generative capacities of computers aesthetically exhausting, and if designers want to use it they should be selective in their choice of outputs, just as the computer can in generative optimisation processes, if programmed well.
Starck is choosey in the presentation of his chair: I didn’t see any images of the hundreds or millions of potential solutions the Autodesk software might have also presented to him. I agree with Ken its not a good chair, not because I have any opinion on its form, but because its an unneccesary injection moulded, mass produced petrochemical piece of crap. Starck’s use of technology is very last century, in that he says he wanted to optimise the manufacturing process. Though, I question that success and I suspect his engagement with the generative process is superficial and its not much more than a marketing ploy by Kartell and Autodesk. I too remember when he came out and said he was retiring because design was destroying the planet (as late as that was in 2008, coming nearly 40 years after Papanek wrote the preface to Design for the Real World). If he had actually retired then, he wouldn’t be tarnishing the legacy of his more interesting work from last century. There are many more exciting uses of generative design in furniture over the last 20 or 30 years, eg. Joris Laarman, Bernard Cache, Greg Lynn and of course Enrica Colabella and Celestino Soddu (nice to see him post to this list).
I would say that the use of generative design in the creation of new products has a very dark potential to continue Modernist-inflected destruction of planetary resources from over production (what Stark spins as ‘democratic design’). So, I think designers should also be selective in their choice of applications. There is greater untapped potential in the use of generative design in creative and transformative forms of repair and reuse, i.e in applications of care and conservation. The generative process is propositionally well suited to the adaptive repair of damage and waste, process which are themselves highly generative given the ranges of how and when products like furniture may break in the real world. Repair, being post-consumptive and often de-centralised in practice, is a better space to limit the power of generative robotic processes in product fields, so as to conserve energy and avoid the over-exploitation of material resources.
best
Guy
Dr Guy Keulemans
Lecturer
UNSW | Art & Design
UNSW AUSTRALIA
Paddington Campus
Cnr Oxford St & Greens Rd,
Paddington, NSW 2021
Phone +61 (2) 8936 0770
Email [log in to unmask]<mailto:[log in to unmask]>
Web guykeulemans.com<http://guykeulemans.com/>
On 15 Apr 2019, at 6:12 pm, Celestino Soddu <[log in to unmask]<mailto:[log in to unmask]>> wrote:
Dear PhD-Design people,
may be of interest my generative design approach using AI in my last paper,
"AI Organic Complexity in Generative Art"
https://www.generativeart.com/GA2018_papers/GA2018_Celestino%20Soddu.pdf
For giving a look to the results of my generative approach to architecture
and design: https://www.generativedesign.com
For me the question is not the optimization process but the possibility to
design our Idea-Vision before the design process, identifying this Vision by
using subjective algorithms. The several possible results, a set of defined
and complex variations, will focus at the best the idea and the possibility
to make recognizable own design vision.
Kindest regards
Celestino
______________
Prof. Celestino Soddu, architect
www.generativedesign.com
www.generativeart.com
www.gasathj.com
-----Messaggio originale-----
From: Sandra Bermudez
Sent: Sunday, April 14, 2019 7:12 PM
To: [log in to unmask]
Subject: Re: Automated design generation and optimisation research
breakthrough
AI and ML powered design services and products relies on what Alan Cooper
points as "working forwards"; making design decisions based on confirmation
of preset data in opposition of "working backwards"; questioning everything
that has been accepted.
https://m.youtube.com/watch?feature=youtu.be&v=f8I7LPb0m64
How might we escape from "perpetuating everything that is wrong with
current reality, ensuring that all possible futures are merely
extrapolations of a dysfunctional present?"
https://www.readingdesign.org/a-larger-reality
On Sat, Apr 13, 2019, 14:14 Krippendorff, Klaus <
[log in to unmask]> wrote:
Dear Terry,
First of all an epistemological fine point: I didn't suggest that
designers "identify unimaginable possibilities." If they are unimaginable
they cannot be acted upon. I said "previously unimaginable possibilities."
Designers need to go beyond both what is commonly understood and hence
imaginable and expand their own space of possibilities.
I grant you that computes can search among an amazing number of data
points to find what someone is looking for, hence imagines, but what is
found must have been there to begin with.
Computers can also explore an amazing number of combinatorial
possibilities, more that humans can combine and examine in their life
time.
But the elements they combine must be finite and known in advance by the
programmers of algorithms. Going through a space of say a billion binary
variable requires 2 to the power of 1,000,000,000 steps, which is
transcomputational. Even with something that is computable, a computer may
come up with unanticipated combinations but not transcend them.
I have a lot of respect for human creativity. Contrary to what you are
claiming, I am suggesting that creative designers are extremely good are
thinking out of the box of deterministic processes.
I am not referring to designers who merely use a more attractive shape or
color of something already known, but designers that propose something
previously unimaginable, develop something not predictable by
extrapolating
past trends, not finding something rarely noticed and increasing its
probability by offering it to a large population.
Klaus
-----Original Message-----
From: PhD-Design - This list is for discussion of PhD studies and related
research in [mailto:[log in to unmask]] On Behalf Of Terence Love
Sent: Saturday, April 13, 2019 6:27 AM
To: [log in to unmask]
Subject: Re: Automated design generation and optimisation research
breakthrough
Dear Klaus,
Thank you for your message.
In it you suggest that one characteristic of design activity is that, '
Design operates in a space of previously unimaginable possibilities.
A problem with human designers is that they are not good at identifying
'unimaginable possibilities'
In contrast, from my experience over the last 45 years, computers can be
much better and more thorough than human designers at finding
'unimaginable
possibilities'. There are many many computer-based approaches. Currently,
in the product design world, generative design methods have since the 80s
been creating optimised designs that human designers have found difficult
or impossible to imagine.
Another example of such generative approaches that go beyond what human
designers can creatively think is the computerised approach that
Phillipe
Stark is now using to extend his creativity - see for example,
https://www.autodesk.com/redshift/philippe-starck-designs/?utm_source=Marketo&utm_medium=Email&utm_campaign=Redshift%20newsletter%20Weekly-2018-10-25T11:00:00.000-07:00%20&mkt_tok=eyJpIjoiTnpGaVpqRTNOamRsWWpneSIsInQiOiJcL3FRd24zOEVtMVwvY0JSTTROdGpBUWc3MW1vYzhHUWMxVmhzYlptSkRsM1ZiR3ZuTzVWSHorWjJiaFQrZkJHK0tad0hCdDBUOVRSRWdOelczRmVjMVwvaThmNmxTNFF5NGUxYnpHVWlWMnJyekZLbjRvNXNSbUhqVnlJV05sQzJnMzQ5RXZnbmVFU0hEUm13c1ZhankxaWc9PSJ9
And... the first full length book machine-written book was published this
week by Springer Nature (
https://aktuelles.uni-frankfurt.de/englisch/first-machine-generated-book-published/
)
Best wishes,
Terry
-----Original Message-----
From: PhD-Design - This list is for discussion of PhD studies and related
research in <[log in to unmask]> On Behalf Of Krippendorff, Klaus
Sent: Saturday, 13 April 2019 1:32 PM
To: [log in to unmask]
Subject: Re: Automated design generation and optimisation research
breakthrough
Hi Terry and Dave,
Yes, computers are faster and can search for optimum solutions provided
the number of alternatives are well-defined and within computational
limits.
Regarding computational limits, design decisions often exceed them which
is why Herbert Simon suggested to replace optimizing by satisficing --
keeping the criteria within computational limits.
Regarding the well-defined mature of computational algorithms, most design
problems aren't of this kind. We have plenty of theories, especially
including mathematical ones, that extract determinisms from the reality of
human interfaces with technology, which may hold for a moment, may be
enforced by authorities but soon become outdated.
Semiotics is a good example. Morris, Pierce, Russell and many other
semioticians abstract meanings from human interactions into triadic
conceptions. Their theories has given rise to algorithms that include
dictionary definitions of meanings -- the General Inquirer, LIWC, CATPAC,
Webcrawlers, Google translator, for example. Such software can mine large
data bases, identify, analyze, evaluate, even vary a lot of phenomena.
They
may create novel texts that may not be obviously distinguishable from what
humans might say -- provided such texts are short and of the most basic
kinds, like Alexis.
Don't forget, by (my) definition:
Design operates in a space of previously unimaginable possibilities.
Good design is inherently opposed to the kind of determinisms that
computers are programmed to follow. Algorithms are routines that when
adopted by human beings keep them entrapped in reproducible habits,
burdened by dreadful conditions, wasteful of their creativity, or
oppressed
by conditions that seem hopeless. On a macro level design is what keeps
culture viable and unpredictable, on an individual level it has to be
revolutionary.
We may not be able to compete with computers quantitatively, but even
everyday design takes place in a space that computers cannot create on
their own.
Klaus
-----Original Message-----
From: PhD-Design - This list is for discussion of PhD studies and related
research in [mailto:[log in to unmask]] On Behalf Of
[log in to unmask]
Sent: Friday, April 12, 2019 11:40 PM
To: [log in to unmask]
Subject: Re: Automated design generation and optimisation research
breakthrough
Hi Terry,
I have been down this route both practically and intellectually in a
number of projects, not with the power of contemporary AI, let alone
quantum computing, but the problems of limits in the nature of computers,
in contrast to people remain. There are various ways of describing the
problem.
My best simple ways of describing it involves a knowledge of basic
concepts in semiotics and linguistics. Charles Morris’s account of
Semantics, Syntactics, and Pragmatics helps. Putting the matter VERY
simply: Computers of any kind, including quantum computers, are extremely
good at precise and fast construction and application of semantic and
syntactic rules. They have NO pragmatics engine. Indeed, as far as I know,
no-one has yet conceptualised, let alone implemented such a system.
Within their limited domain computers can do great things. They can also
make bigger mistakes faster than any person could possibly even conceive
of. Our ultimate defence, at least for the moment, is to know where the
plug is so that we can pull it out.
David
-----------------------------------------------------------------
PhD-Design mailing list <[log in to unmask]> Discussion of PhD
studies and related research in Design Subscribe or Unsubscribe at
https://www.jiscmail.ac.uk/phd-design
-----------------------------------------------------------------
-----------------------------------------------------------------
PhD-Design mailing list <[log in to unmask]> Discussion of PhD
studies and related research in Design Subscribe or Unsubscribe at
https://www.jiscmail.ac.uk/phd-design
-----------------------------------------------------------------
-----------------------------------------------------------------
PhD-Design mailing list <[log in to unmask]> Discussion of PhD
studies and related research in Design Subscribe or Unsubscribe at
https://www.jiscmail.ac.uk/phd-design
-----------------------------------------------------------------
-----------------------------------------------------------------
PhD-Design mailing list <[log in to unmask]>
Discussion of PhD studies and related research in Design
Subscribe or Unsubscribe at https://www.jiscmail.ac.uk/phd-design
-----------------------------------------------------------------
-----------------------------------------------------------------
PhD-Design mailing list <[log in to unmask]>
Discussion of PhD studies and related research in Design
Subscribe or Unsubscribe at https://www.jiscmail.ac.uk/phd-design
-----------------------------------------------------------------
-----------------------------------------------------------------
PhD-Design mailing list <[log in to unmask]>
Discussion of PhD studies and related research in Design
Subscribe or Unsubscribe at https://www.jiscmail.ac.uk/phd-design
-----------------------------------------------------------------
-----------------------------------------------------------------
PhD-Design mailing list <[log in to unmask]>
Discussion of PhD studies and related research in Design
Subscribe or Unsubscribe at https://www.jiscmail.ac.uk/phd-design
-----------------------------------------------------------------
|
