Atomic Design, AI, and Robot Overlords
Updated: Mar 21, 2018
I've been hearing a lot of discussion lately about Brad Frost's very cool Atomic Design book and project. Although it is of no fault at all to Atomic Design, I feel of two minds about this. All of this classification and automatic size adjusting of the systems is great, and makes good sense. The hesitation that I feel in accepting this system is that it is a stepping stone on the path leading us slowly into systems that design themselves and are automatically responsive without human guidance. Indeed, AI is the future regardless of how we feel about it. It is as inevitable as the steam engine. Which is part good, and partly leaving me with a sense of nostalgia. We are getting closer to self driving cars, what about self-adjusting design systems? At what point will drivers and designers no longer be necessary, or will that point never come?
Semi-Automatic Design System/Atomic Design Pros:
A. Futuristic, forward looking in that it lays the groundwork to automate tasks
B. Time saving in that it has coherent structure and consistency in testing
C. Long-term money saving in employment cost through automation
D. Money saving by having the potential to automatically respond and adapt to which elements get the best rates (e.g. highest purchase levels, lowest bounce, longest read times, etc.)
E. These principles are being used to a certain extent in writing with automated journalism which while debatable morally is undoubtedly profitable
F. Applying a consistent taxonomy to web design is a good thing to do. Also we come up with such delightful names. Like the "Meercat" ad that pops up from the bottom and "Hamburger buttons." I haven't found a UX Dictionary yet. If you know of one shoot me an email.
G. It's a bit of a hail mary for coding, with a library to build from and thinking of things in terms of flexible elements instead of static pixels.
H. Sometimes AI can pick up bad habits. Like Microsoft's chat bot that Twitter taught to talk like a Nazi.
A lot of this fits into the direction of material design, AI, and makes sense. Here are the things that give me pause:
Semi-Automatic Design System/Atomic Design cons:
A. As a biology lover I'm not fully on board with the Biologically influenced classification system. There is too much baggage in my brain to repurpose this concept simply (are they animal or plant cells? Are they waxy? Where is the mitochondria?)
B. One of the tenants of UX Design is allowing an aesthetic to be grown and curated organically by way of user testing. It seems like the automated nature of this necessitates a "one size fits all" approach to design where organic insights cannot happen naturally. It's the thousand monkeys at a thousand typewriters approach to problem solving.
C. A parallel to this could be seen in the frontier of investing, with a lot of new start ups writing programs that predict stock prices and buy and sell en masse based on their algorithms. These automatic purchasing systems come with their own sets of advantages and drawbacks. I predict similar issues paralleling automatic stock market investing with automated design systems. There could potentially be a David and Goliath effect, where the little guy is more nimble.
D. People adapt. When an ad is consistently featured in a similar location, people learn how to avoid it. When a fashion trend fully takes over, it is no longer cool. Subtle signifiers like political affiliation and region play into what people feel makes a design good, it is subjective and constantly shifting. Design is part of how humans discern the in crowd from the out crowd, what is fresh and what is belonging. Will the design program be able to change in a way that can account for these constant shifts in the way a human designer can? Maybe, but it will have to be one hell of a program.
It seems that regardless of how designers feel about it, this is the future. Best to ride the wave of new technology, keep adapting, and try to steer the boat in the right direction.