A fear-driven UX industry

What AI actually does, and why you’ve been told to panic

I think I can speak for many designers here, that we are in fact afraid of what the future holds for our industry and the perceived value of the service we provide as user experience (UX) designers.

I honestly didn’t think an article like this was still necessary this far along after the popularization of generative AI tools. Because, here I thought that most designers would have tried them and then very quickly realized their limits and the capacity of the tools’ training. I thought more people would be interested in reading about the true capabilities of AI. To learn the limits and to learn about what it can or can’t actually do.

However, here we are, years later, and the marketing hype around the AI industry has fooled many people into thinking it is much more capable of a tool than it can foreseeably be. Especially when it comes to talking about replacing designers.

As it currently stands AI is a tool of prediction, statistics, and math. Nothing more.

What that means is that AI does not think in the way you do (source), it does not have critical reflection, and it does not feel emotion, only recognizes the pattern of it. It does not understand the meaning of what it produces (source). AI is still a black box. No one truly understands just yet why it comes up with some of the things that it does. Or how it comes up with it. And the more we train them the less reliable these tools become:

“early models often avoid user questions but scaled-up, shaped-up models tend to give an apparently sensible yet wrong answer much more often” Source: https://doi.org/10.1038/s41586-024-07930-y

AI tools are hallucinating more frequently, with more confidence, and are less reliable than earlier models. All that said, most people I interact with on the subject don’t talk about reliability or efficiency, but rather they bring up their fears for the future of the design industry, the future of their jobs and their careers.

Design is being de-valued, because anyone can go and type in a few words and come up with a prototype, design, image, or body of text that on first scan looks half decent. However, as you dig into it you realize that everything is a little off, the styling is incorrect, the arrangement is strange, or it completely misses the mark. This comes back to people understanding the work that we actually do as designers. I feel like this is a problem that I am continually working through, because no one questions when a design is good, when an experience was good. We all know when it’s bad, when something is so confusing that we give up trying to use it. But we categorize those things as drop off rates or low conversion. We don’t consider the impact a UX designer can have on making that flow make sense. We just think of designers as someone who works on visual layout rather than the person who understands the psychological and behavioural principles behind why a design or layout just makes sense.

We also continually read and hear that we will be left behind if we do not adopt AI tools. “AI will not replace designers, a designer using AI will” comes up over and over in my feed.

In some respects I consider this statement bullshit. This comes down to whom we are as critical thinkers in the field of design. It comes down to how we were taught to design; the principles we know and understand as designers. AI tools aren’t what make us good at what we do. This is a marketing ploy to make more designers jump on board and use these tools, to become reliant on them, and pay for them.

Studies are showing that Generative AI does not necessarily help us be creative, rather it may end up making us less creative over time.

“overdependence may compromise creativity, critical analysis, and academic integrity…In other words, technologies like ChatGPT may make some research jobs go faster, but relying too much on them might hurt the quality and depth of the work.” source.

Table 7 sourced: https://www.mdpi.com/2227-7080/13/11/486

AI is only good at disseminating and finding patterns in large data sets. Meaning that while it is a very good tool of prediction and pattern recognition, it is best at giving you a synthesized or generalized output. In fact it can be a really good tool if you just want to know what the most likely or generalized solution to a problem is. I find it useful if I’m struggling to think of what might be expected or the norm in a specific style of document layout. But that’s kind of the limit. If I want it to lay out a specific wireframe or document, I always find it falls short and I end up redoing the whole thing anyways.

To bring this back to another example, I’d like to relate this to the invention of the computer and how that disrupted the graphic design industry. For the most part, all of us use computers to design nowadays. However, we aren’t good at design because we work with a computer. Sure I can use Photoshop, but that doesn’t make me a good photographer. And our classic water cooler example: anyone with a Canva license thinks they are a designer—but you know the difference.

People are mixing up the ability to use a tool with the years of knowledge and training and inherent talent that goes into becoming a professional in a field.

AI is similar. There are many pre-computer design skills that are highly valued still today, knowing some of these skills can help you lead in the field. Some of the best graphic designers I know, still paint by hand, lay out posters physically, or stage elaborate photo sets to achieve something that truly stands out.

[embed]How Jessica Walsh Developed Her Distinct + Colorful Style of Art Direction at Print Walsh tells us the story of how she developed her distinctive early style…eyeondesign.aiga.org

These skills will only become more valued. Because, AI is really only good at creating mediocrity.

This is because to train an AI you need a vast base of source data for it to learn from. Meaning that specialized knowledge is almost impossible because there is not very much source material to learn from. It will undoubtedly fail at any prompt asking it to do such kind of work, because what we do is highly skilled, creative work—the highest level of work that automation can only dream to approach or replace.

[embed]The Robot Curve - MARTY NEUMEIER by Marty Neumeierwww.martyneumeier.com

The money isn’t in replacing designers (we aren’t the part of the production process that costs a lot of money), but rather in the lower curve area of repetitive work.

AI cannot make everything better, we want AI to work because what we do is hard, building the future is hard.

But when was building something worth our effort ever easy?

“All things excellent are as difficult as they are rare” —Spinoza