Algorithm-Assisted Design?

(Introducing, your new principal designer!)

The design field is in a very odd state. At least on Twitter, we seem to always be in a constant state of disagreement. But one thing I think we can all agree on, is that we are all utterly disposable, and quite possibly will one day (maybe soon) be replaced by algorithms.

Design vs. Production

I have often said that there is ALWAYS 2 classes of work for people in the design field: Design, and Production.

“Design” is the creative side, the exploratory side. It’s the conceptual thinking side that every designer worth their salt has dipped their toes into its murky depths more than once in their career. It craves research and understanding. It craves stimulus and introspection. It is the kind of thing that can be both enlightening and terrifying at the same time.

“Production” on the other hand, is concrete and diligent. It’s consistent and easy. It’s planned. It’s scheduled. It’s safe. It’s not going to make your mom sad. It’ll make your dad brag about you at the family dinner table. It will make your CEO smile.

We all sort of understand that people can do both, all at once, but not without accepting certain compromises. But I’ve sort of gone off topic a bit here.

The point is that not every designer does things in precisely the same way. Some folks are more “production-minded”, and others are more “conceptual”. With the vast number of companies in tech, big and small, many of these organizations are building production teams to be able to ship digital products as fast as possible. Here is where it gets scary.

I wrote a long Twitter thread about how the design industry may see a big downsizing in the coming years. The idea is that if production work is basically just intelligently selecting UI patterns and components to implement, we can already train algorithms to do this for us, and therefore eliminate, in one fell swoop, approximately 50% of what the Design industry provides to companies, and it may not be long until computers do everything else as well.

Today, we know algorithms are already deployed across the tech industry’s biggest services. Google, Facebook, Twitter, and more (?) have deployed machine learning as an integral part of the product, which significantly shapes our day-to-day experiences. Especially in light of all the recent election scandals, we know these sophisticated algorithms can be used to manipulate people’s thinking and influence their life choices. So it wouldn’t be that big of a stretch to imagine that algorithms can also make intelligent decisions when it comes to UI production.

Either knowingly or not, we seem to be marching in a direction that is really well-suited for training algorithms to do production work. If we can distill design decisions into a series of mechanical choices based on data, then there’s no reason that humans would need to be involved at all, after a certain point.

Image result for machine learning scatter plot

(source) This is fundamentally what design decisions might look like to a machine.

Algorithms need data and contextual factors to train their decision-making abilities. Contextual factors can include many of the different, real-world facets of design. Things like typefaces, layouts, screen orientations, viewport dimensions, we can pretty much encode all “best practices” in terms of scatter plots (like the one above) that an algorithm can understand.

Going a bit further, if we encode the algorithm to respect Jakob Nielsen’s 10 Usability Heuristics, combined with the teachings of many great designers of the past (their legacy is “encoded” in design books), then it probably wouldn’t be too much longer after that algorithms can start making decisions on the creative, conceptual side of things as well.

Countering the Doom

So, I mean, it’s not all bad. One of the things I felt early on was that having algorithms do a lot of the production work, can be a good thing. Having computers take over some of the less interesting aspects of our jobs is something I would look forward to… One way in which we should consider how this works is by thinking about something I’m going to call algorithm-assisted design.

Autodesk has been exploring this topic already.

With the idea of generative design, Autodesk has basically created an application that allows someone to input the various parameters (also known as requirements) of a particular design problem, and the software generates multiple variations of things based on the encoded parameters. The “designer” can then review the variations and adjust parameters to generate even more ideas or select different options to take to the next stage.

Now, I don’t know about you, but the idea of making variations is such a turn off to me, that I wouldn’t mind having an algorithm do that for me.

A lot of what Autodesk’s generative design software does is largely for physical products, because they make a lot of the CAD software that allows people to create physical goods. Once you have the variations you want, you can then take them directly to a CNC machine or a 3D printer to product something tangible, to see if it would actually work as intended.

This same kind of process can work for purely digital design. We can input parameters that the computer needs to generate variations, look over the results, and select the ones we prefer.

In this worldview, algorithms aren’t necessarily replacing humans, but collaborating along side them, and in the best situation, coming up with new ideas that people wouldn’t have thought of all on their own.

Where will we be in 30 years?

Well to wrap things up, I want to hope for the positive, that we aren’t inadvertently making our own industry completely and utterly irrelevant. My concern is that we are headed in that direction at the moment, but it has always been my feeling that human ingenuity is something that isn’t easily replaced.

In 30 years, I would like to imagine that we could do away with a lot of the existing (traditional) thinking about how society can work. Algorithms can quite possibly replace or eradicate all kinds of different jobs that are currently occupied by people. The tech utopian view is that these people can then be freed up to do exactly the things that they’re most passionate about, since they wouldn’t have to waste time on mundane things that algorithms can do already.

Image result for star trek tng starfleet academy

Lt. Commander Data’s image at the top of the essay wasn’t just a funny association. Star Trek already envisioned a society that is essentially a tech utopia. Realistically, things probably won’t look exactly as they imagined on this TV show, but it’s nice to think that this is along one possible future for us as a species. The only question is whether or not we can ever truly get there.


Thanks for reading! If you liked this essay, please share it! If have any further thoughts about what I wrote, feel free to reach out to me on Twitter!

Algorithm-Assisted Design?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s