UX designers automated by AI? When will it happen?
By Tobias van Schneider Published August 17, 2020
A few weeks ago, I posted a tweet, based on another tweet. It wasn't *that* controversial, but enough to get the attention of a couple UX designers (frankly, UX designers are easy to bait). Think of this as an extension of my tweet.
In the tweet I quoted, you can see a person building a little tool based on the GPT-3 AI model (never heard of GPT-3? Start here), which then proceeds to automatically "design" an Instagram looking-app based on his "natural language" description.
Let's not fool ourselves. It's a quick experiment. A proof of concept, nothing more. This plugin alone could hardly replace anything or anyone. But we have to look a bit further than that. AI is still progressing, but it is happening fast enough for us to ask some serious questions.
A big one being: Can AI replace the work of a UX designer?
When I talk to other UX designers about the subject (I'm one myself, just not so religiously attached to the title) I'm met with unwavering confidence that nobody will ever replace their profession.
But I wouldn't be so sure about it. If I can guess, about 80% of what UX designers do today will be automated in the coming 5-10 years. Some designers will be able to adapt just fine, and some will struggle. It's the same thing in most industries, whether we like it or not. We chose the race against technology.
Let’s look at some of the arguments posed in these conversations:
"UX designers can't be automated because the work is creative by nature, and only humans can do it."
Let's be honest here for a second. Yes, this is true for *some* design, but the majority of work that is put out these days is mediocre at best. And I'm not trying to be negative here. Mediocre is what's being asked for.
For example: The majority of apps you see these days work pretty much like any other app. They're all based on Apple's or Google's "Human Interface Guidelines.” They all follow the same best practice UX flows. They all follow the same UI patterns and UX paradigms.
Both UX and UI designers have been working hard to standardize not only patterns and systems, but also libraries and guidelines. In a way, we've been automating ourselves out of our jobs for the last couple years already. It's why you can take a 3-month online course and become a certified UX designer. There's barely any other profession *that* easy to enter.
Giving all these pieces to an AI and letting it figure out the rest seems to be a trivial exercise. There's no question about it.
Of course, true creativity for an AI might be some time away. But true creativity isn't really what's needed for the *majority* of apps or websites these days, at least by the look of it. Most look and work the same anyway. Who are we kidding?
"Yes, but UX design is so much more. It's about user research, empathy and data – only humans can analyze it in a meaningful or effective way."
This is the argument I find the most entertaining because, if there is *anything* that can beat humans with research, pattern recognition and analyzing data, then it is most definitely a machine.
In fact, I'd even argue that an AI can know *a lot* more about human nature than we know about ourselves. If you think about it, we’ve taken this for granted since the advent of computers. Most of our design decisions these days are data-driven. They're rarely "human" driven (although I am still very much a believer in that). But the tech industry has made it a habit to drive every single decision on hard data.
If the data says to do it this way, we do it this way. I would even argue, at most large companies, UX designers do nothing else but designing A/B tests. Throw them against the machine, let the machine decide, and then implement it. UX designers have become glorified servants of the data.
But you know who could design even more A/B tests, and do so even faster and more efficiently? Artificial Intelligence.
"OK Tobias, why are you trying to ruin UX designers?"
I'm not. Nor am I trying to ruin part of my own profession. But I have to look beyond the pretense and be honest with myself. I like to ask questions and see where the answers lead me.
We still don't know exactly what tools we will build to leverage AI in a way that can become "dangerous" for us. Right now, AI is just a thing. It's a playground that can do a lot of things, but nothing really well. But give it five to 10 years and things will look quite different.
I don't expect to have the same job I have now in five years. Things will change, most likely in a direction I least expected. Asking myself uncomfortable questions prepares me to some degree. I expect nothing and everything. I do expect uncertainty.
But what I ask myself is:
Are we doing ourselves a disservice by settling for mediocre design, just because it seems to "work better?"
Do we really benefit by automating our craft with systems, best practices and guidelines?
If all of our work is only focused on "improving the metrics,” then how could we possibly compete with an AI?
I don't know, but we will find out.
P.S. On a slightly different note, here's a little snippet from a chat I had with the GPT-3 AI. I found it interesting and quite beautiful, in a way. It even made a typo, cute.
Hi, I'm Tobias, a German designer living in New York. I'm the author of this blog, nice to meet you!