Well, shit, Figma. Nice one lads. Switch to Penpot by the way.
With the recent announcement of Figma’s AI feature-set at the corporate vibe-fest that is Config, there’s been something of a clamour to opine, positively and negatively, on the impact these features will have on designers and the industry as a whole. Now, a lot of these are actually half decent, minor features that might save us a few minutes here and there. As a serial ‘Group 24098’ layer-name-forgetter, the ‘rename layers’ functionality is sound. The enriched search even veers on actually useful.
However, it’s the ‘make’ tooling that most folks will be cursing, or celebrating, or rushing to make TikToks about because that’s apparently how design advice is consumed these days. Firstly, labelling it ‘make’ instead of ‘generate some random fucking slop’ or whatever else it was called previously is a genius marketing move, pretty shitty in every other sense, but the content designers should get a raise for that one.
They’ve nailed the typical hype of every other startup and scale-up throwing an AI hail Mary into their product too: showcased a best-possible-scenario demo at a heavily curated event to an audience that would be mostly receptive to it, or at least politely clap, because they can pretend they’re part of a community while their job is commoditised in front of their very eyes. In its current state, the ‘make’ AI feature looks rather simplistic, with a bunch of pre-designed components powering the prompt outputs. In the future, it’s transparently clear that Figma want to train their models on real projects, especially with the shameless ‘on by default’ AI-mining setting for starter and pro plans.
I want to be very clear that I hate this. I hate that Figma have gone down this route, I hate that they’re acting like every other corporate entity once they corner a market and see themselves as a future monolith. I hate the fact that Adobe shitting the bed on their deal to buy them gave Figma a second chance to not be dickheads and they doubled down on the dickheadery. I hate the fact that I’m disappointed and even surprised that a corporation is acting like a corporation. But, is AI going to do all our UI design for us now? Should we be worried? Should we care?
As much as I’m a doomer about AI (and Figma now, after their CEO outing himself as the biggest gobshite in design and their incessant pursuit of cornering the design tools market) I want to present something of an optimistic take on this.
You might know me as a big cynical fuckwit, but honestly, I am a happy and cheery fella! And I want nice humans to do well.
Bullshit in, bullshit out
The average designer is… well, average. That’s not a bad thing. We can always get better. I’ll be the first to admit that my visual design skills are somewhere between ‘spartan and functional’ and ‘absolutely unhinged’. Take a look at the vast majority of community files in Figma and you’ll see that the general quality level is what you’d expect from somewhere between junior and senior designer. Again, no slight on these people, seeing designers grow and develop their craft is heartening and joyous for me, and designing in the open, sharing your work, and letting others learn from it is one of the fastest ways to grow and network as a designer.
That doesn’t really change the fact that Figma’s AI shit will suffer from the same problems every other company’s GenAI shit suffers from: the average input to its dataset is, almost by definition, mediocre and derivative. Especially when you consider the state we’re in by and large as an industry.
Everything looks the fucking same
We’re at a bit of a crisis point of creativity in UI design. Every startup ever seems to have some clone of the Linear site at the top of their marketing funnel. Every product has been homogenised to the point where you need only switch out a logo and a primary colour to make it look almost exactly like its competitors. Indentikit design systems and UI libraries are used almost without question. It’s a frankly quite shitty time to be a UI designer, because we’ve moved past trends and well into the realm of cheap dupes. Clients want this, bosses expect it, and we’ve seemingly built a rather nasty habit of correlating ‘quality’ with ‘how close this looks to every other fucking thing that’s being shipped right now’.
So across the industry we’re shipping low-effort copycat shit. And now Figma have the good grace to train their AI on some/all/who the fuck knows what combination of this low-effort copycat shit, to then poorly generate a visual Markov chain, through machine learning designed to churn out even lower-effort, copycat shit.
You shouldn’t be worried about what this says about you as a designer. You should be worried that over-eager non-design people equate this work to ‘real’ design; design done by humans, who understand other humans, who imbue their work with the craft and emotion that only humans can express. It’s a problem of perception as much as it is a problem of output. Good job we’re all already well experienced at trying to justify design to someone who signs checks but has absolutely no clue what design is, or where their yoghurt is supposed to go when they eat their packed lunch (it’s your mouth, Callum, you dense prick).
We’ve been here before though, and it wasn’t because of AI.
Design is already commoditised
Go on Fiverr, or any other ‘cheap’ freelance marketplace, and you’ll see people offering logos for a few dollars, site designs for a couple hundred, full builds for less than the average day rate. These are almost exclusively cookie-cutter projects that give clients and stakeholders an output without any accountability or commitment to outcomes. Maybe that has its place, but it’s far beyond the scenarios that we should be aspiring to involve ourselves in as designers who give a fuck about other people.
I’ve decided I’m going to have the same response to the AI-doers as I’ll have to prospects and stakeholders who tell me ‘ChirpyLogos412 can do this for $69 and you charge more than that an hour’ which is to ignore them, wish them well, or tell them to go play in traffic, depending on how good my morning coffee was.
That’s because I know the value good design can bring to a project, and it’s not the output. Design is about humans, about sense-making, systems thinking, and craft. It’s as much a process of discovery and decision-making as it is a deep work. ChirpyLogos412 doesn’t give a fuck about the people who use the thing they make for you, doesn’t give a fuck about your business goals, your desired outcomes, about accessibility and the real-world impact of their work. And neither does this new wave of AI-generated UI slop.
Staying ahead of the slop
So where does this leave us? Firstly, the problem of perception that’s plagued design for a long while is only going to be exacerbated by the facade of ‘magic’ behind which GenAI conceals itself. The same people who wanted to spend $69 on a website will now want AI to do it for free. Fuck them, let them, they’re not your people and they never will be.
Next, and this especially true for newer designers, you might find yourself feeling like your work is ‘worse’ than what is being generated by the various AI tools out there. That’s fine. You can get better. Learning is an intrinsically rewarding and fulfilling experience, and it’s okay to be bad at shit. Everyone sucks shit at something until they don’t. The learning experience is something uniquely human, improvement and mastery is integral to self-determination. Stick with it.
Finally, we need to revisit our craft. What makes us unique as humans? As individuals? How do we imbue our work with this? It feels like a lifetime ago, but there used to be so much room for expression and exploration within UI and product design. There was equal room for minimalists who focused on excruciatingly incredible typography and meticulous grids, maximalists who lived and died by the vibe, pragmatists who focused on core principles, and a thousand different sub-styles in between. Yes, this is an overly-romantic, nostalgic reflection, but design before everything congealed into a mush of airy, Inter-backed, indentikit shite was actually fun, and you’d go to different people for different things.
My bet is that we’ll see similar with this emergence of AI generating milquetoast, mediocre shit. ‘Designed by humans’ will become a selling point. People will be hired for what the robots can’t do. We can fight back by focusing on what makes us human, what unique perspectives and practices we can bring to a design process, and by just having some fucking fun again. Bring back the weird web.
Being human
For me, I’m focusing on a few things that I’m pretty confident some cobbled-together AI screens won’t be good at, or capable of, for at least a while. I’d like you to do the same. Grab a pen and paper or your favourite AI-generated note taking app and follow along with me as we explore A List Of Very Human Traits That Means We Can Say Fuck You To The Capitalism Robots.
Firstly, systems and product thinking. I’ve worked a tonne with early stage startups and these are the elements that keep getting me hired. One of my biggest skills is making sense of the chaos that exists in the early stages of ideas, where everything is just vibes and chaos. I can help founders express their ideas, I can help teams explore and understand a problem space, I can connect dots, visualise system models, translate them into designed models, design, test, and iterate. I can also hold myself accountable to hypotheses and help objectively evaluate others’. This is a very basic UX or Product Design skillset, but I’ve done it so much, and iterated on it every time, that I’m very good at it. I even wrote a book on it.
Secondly, I can code. I write very solid HTML and CSS, I am decent with JS. I can build bootstrapped products or early iterations from scratch, I can write APIs, I can code prototypes, I can handover CodePens to engineers to express my ideas in the medium in which they will eventually live. Being able to code my own work means my design process for projects where I’ll do so looks very different, and my design files look like a serial killer tried to do a SuperHi course. This means that not only can I act with volition and shape my process based on the project, but that handover is more straightforward, my work is realistic, and I understand enough about how browsers work to push for very important things like accessibility and performance without nagging or hoping someone else does that work.
Finally, I make weird shit. I like sites and products that subvert the status quo. I like being able to fuck around and innovate in the areas that matter. I know the difference between convention and copying. I love the weird web. I think SaaS has given us all brainworms that make us approach every project as if it’s a future unicorn. I like pointless fun features, personal touches, and weird visual design. As AI brings about an inevitable convergence, I’m all-in on responsible and creative divergence.
Over to you
What’s your list? What separates you from the scary design robots? Maybe you’ve got an eye for brand and illustration that’s unequivocally ‘yours’. Maybe you’re an incredible researcher and early stage designer who knows what it means to democratise systems. Maybe you work away from the noise of SaaS on actual real problems instead of some rich cunt’s vanity project or VC-backed behemoths. Maybe you can code. Maybe you’re just incredibly fucking good at your craft.
Remember: the floor might have been raised by this shit, but we’re responsible for raising the ceiling. The best that we can do will and always should outpace the average of what a bullshit Markov chain in a beret can do. This is a chance to move to the margins and start differentiating yourself. Make ‘designed by humans’ something aspirational.
Oh, and remember to turn those AI features off. You shouldn’t be paying Figma with your work and ideas.
