Reply

A.I. "designed" cars

December 20th, 2018, 9:38 am

AVClub
step three
step three
Posts: 191
Joined: November 22nd, 2014, 9:32 pm
Check it out, A.I. is now "designing" cars using a technique called style mixing. Combine this with Nissan's desire to create a complete A.I. driven designer I would say maybe a good amount of design jobs are not totally safe from automation in the near future after all. Would be great to have a discussion and learn how some of you feel you will adapt your skills, or maybe you don't think that's even necessary.

https://jalopnik.com/these-ai-generated ... 1831176960

Re: A.I. "designed" cars

December 21st, 2018, 1:24 pm

User avatar
ralphzoontjens
full self-realization
full self-realization
Posts: 951
Joined: February 3rd, 2010, 10:20 am
Coroflot: 76078
Location: Tilburg, the Netherlands
Yes, I am learning to program that A.I. :) Also with a coupling towards manufacturing and developing knowledge on user requirements regarding customizers and generators.
http://www.id-z.one
IDZone - Product Design || Visualisation || 3D Printing

Re: A.I. "designed" cars

December 26th, 2018, 8:50 am

AVClub
step three
step three
Posts: 191
Joined: November 22nd, 2014, 9:32 pm
ralphzoontjens wrote:
December 21st, 2018, 1:24 pm
Yes, I am learning to program that A.I. :) Also with a coupling towards manufacturing and developing knowledge on user requirements regarding customizers and generators.
Interesting! What are your thoughts on how this will impact the Industrial Designs field? Wipe us out completely? A tool that is our new best friend?

Re: A.I. "designed" cars

December 26th, 2018, 9:48 pm

User avatar
Cyberdemon
full self-realization
full self-realization
Posts: 3442
Joined: February 7th, 2006, 11:51 pm
Location: New York
IMO, AI will continue to replace tasks which meet two conditions:

1 - They can be described as having a solution which is "good" or "bad". (A picture of a face can be objectively considered good or bad, as can the design of a product).

AND

2 - The costs associated with the human labor of creating or calculating that solution warrants substantial enough development of the software tools.

This is where I don't see designers being replaced any time short term. Why? Good design is often polarizing, complex, and built to address hundreds of underlying characteristics such as the ideal human grip, the underlying technological stack, the functional properties of its material, the costs of those materials and steps needed in their unique tooling, as well as a pure aesthetic which many may consider bad.

AI could probably tackle any of those areas individually, and over time begin to combine their functionality. This is where you see FEA tools hitting their stride in optimizing for a set of inputs. But the human aspects of that are much harder. How do you quantify aesthetics vs drop performance vs tooling costs? A human would have to review not only a CAD design but a BOM, cost implications, tooling implications, and now you aren't just replacing a designer, you're replacing the entire engineering division. And if one of those AI Designs solution fails in the field and the AI designed bus kills 60 people? No one would ever use it again.

For reasons like that, most of the work I think will continue to focus on industries that spend a ton of money on shiny new things like media and entertainment. If someone can make an even more spectacular super hero movie to replace 100 animators, their risk of failure is very low. Or AI generating an entire video game world on the fly (which has already been done in games, but will continue to get more and more complex).

I would say for anyone whos already graduated college, your careers will be safe, and any AI tools that do emerge will just become part of your workflow until you retire. No one is going to show you to the door because the new mainframe has taken your job.

I can't say the same for factory workers.

Re: A.I. "designed" cars

December 27th, 2018, 11:25 am

User avatar
yo
Administration
Administration
Posts: 17355
Joined: January 5th, 2004, 6:57 pm
Coroflot: 67242
Location: SoCal
Predicting the future is always a tricky thing. We rewatched "2001, A Space Odyssey" on Xmas (so nice to watch it now that I'm older on a big screen and with big speakers! Last time I saw it was on VHS). It is amazing what they got right and wrong in that movie. For example the astronauts were using what looked very much like iPads... predicted in 1968, way before technologies like flat panel display, powerful microprocessors, WiFi would be popularized.... however those iPads were labeled IBM, and when they have to go and shut down HAL (spoiler alert) the main computer still takes up an entire room sized structure (instead of being the size of a phone).

My point being, will this happen? Maybe. It is one extreme outcome. More likely we will have a host of new tools to make out jobs easier.

I think the disruption facing car design isn't if computers will design them, it is if people will buy them. If the majority of them become shared devices, what are the implications for design and how many models/brands will be needed? Will design become more use case focused and less focussed on personal and brand expression?

This kind of disruption could turn that industry upside down faster. Look at how fast Uber and Lyft changed the taxi landscape. I am very surprised that car makers have not launched "ride share" editions with more durable interior materials, beefed up suspensions, longer lasting transmissions and the like. It shows how slow that industry is.

I think the most important thing is to remain flexible and open while knowing what your value is and how you can contribute.

Re: A.I. "designed" cars

December 27th, 2018, 2:56 pm

User avatar
KenoLeon
full self-realization
full self-realization
Posts: 702
Joined: October 23rd, 2012, 12:22 pm
Location: Mexico City
I spent most of 2018 working on A.I. related projects, the first one was working with a startup/ hedge fund that is trying to replace traditional hedge funds with crowdsourced A.I. ( machine learning algorithms ) and the second a self published book on C.A.I. ( Conscious Artificial Intelligence) this is something that I’ve been researching for close to a decade and still a very new and mostly theoretical field.

What I gather is that we are still in the phase where A.I. needs humans to provide the data and define the problem,plus actually plan and code the A.I. but increasingly the A.I. tools available are getting generalized & democratized, so many industries will need to adapt and some disruption will happen, a hedge fund manager needs to know about A.I. quant strategies and a design/engineering team can use A.I. tools as complementary for a number of roles ( ideation, data science, parametrization,optimization,etc,etc). Some will invariably try to substitute the hedge fund manager and the design team for cost cutting and novelty, but that would be their bet to make.

A bigger disruption and novel design and production ways might come when we achieve and somehow commercialize a general or conscious A.I. ( I argue in my book that it is mostly a semantics, scale, systems problem), but I believe we are still a few decades away.

A designers job security confronted with the current AI landscape is hard to predict, but I see it as an opportunity rather than a challenger for now, in my day to day it is design that helps the A.I. rather than the other way around.
Eugenio (Keno) Leon
k3no.com Linkedin Instagram

"Go where you are celebrated, not merely tolerated"

Re: A.I. "designed" cars

December 27th, 2018, 5:53 pm

User avatar
yo
Administration
Administration
Posts: 17355
Joined: January 5th, 2004, 6:57 pm
Coroflot: 67242
Location: SoCal
Keno, I think that is a good framing of it. Some opportunities might go away, others might open up. A big chunk of what design as a general field is moving to is big picture consultative type engagements on the out of house side, creative direction and stewardship on the on house side, and very nuts and bolts creative solutions in design for manufacture on both in and out of house. A lot of the tasks in between those three poles could possibly be automated or vastly streamlined through automation... It already has been. What an individual can do now in a few days used to take a team of people a month 3 decades ago. Maybe that same task took small group a week or two 15 years ago. AI or not, OI think the tools will continue to make that true.

Assuming this may be tru, there will be pros and cons. Possibly less jobs, especially less entry level jobs or the kind of jobs where an employer needs an extra set of hands. What will that mean for those of us who feel an obligation to mentor the next generation? On the flip side, there may be more opportunities perhaps for those with experience bringing things to production or who can quickly synthesizes a lot of seemingly unrelated data points into cohesive ideas and strategies. How will a business full of experiences (expensive) partners be able to scale?

Re: A.I. "designed" cars

January 3rd, 2019, 5:54 pm

User avatar
yo
Administration
Administration
Posts: 17355
Joined: January 5th, 2004, 6:57 pm
Coroflot: 67242
Location: SoCal
"Forbes writers will use AI to pen their rough drafts"
https://venturebeat.com/2019/01/03/forb ... gh-drafts/

Re: A.I. "designed" cars

January 4th, 2019, 8:05 am

User avatar
Mr-914
Moderator
Moderator
Posts: 5832
Joined: January 12th, 2004, 7:44 am
Location: Montréal, QC
It's interesting to note how derivative all those cars are. That's obviously how this product was designed, but it's not what makes successful product design as the top selling cars and trucks of 2018 do not look like they are from 2008. Sure, the form factor is traditional, but the exectution is creative. I can see AI helping us come up with ideas, but I think it will always be someone's squishy wetware that will use their human judgement and creativity to decide the direction and refine the details.
Ray Jepson | Twitter | LinkedIn

"The key to success in this business is to find a boss who doesn't care." - Mike Rowe

Re: A.I. "designed" cars

January 4th, 2019, 2:35 pm

bcpid
step four
step four
Posts: 397
Joined: March 20th, 2004, 8:26 pm
I think this is fascinating, and as presented, it applies only to styling which is great. The other, interesting 99% of design - framing the problem and exploring conceptual paths forward - is still untouched for now. But how cool would it be to plug in visual language rules and see 1000 permutations without lifting up a pen?

Re: A.I. "designed" cars

January 5th, 2019, 7:58 am

User avatar
ralphzoontjens
full self-realization
full self-realization
Posts: 951
Joined: February 3rd, 2010, 10:20 am
Coroflot: 76078
Location: Tilburg, the Netherlands
I see it like that too, an insanely advanced tool. But we will never be able to appreciate an AI 'masterpiece' as some people can create, we are at least very far from that point. A computer simply does not have the comprehensiveness to fully understand what they are doing and be conscious of their motivation behind it.
http://www.id-z.one
IDZone - Product Design || Visualisation || 3D Printing
Reply