Is AI replacing UX designers? (The reality no one talks about) | Bruno Figueiredo

Is AI redefining UX design or replacing it?

In this episode of Human × Intelligent, Madalena Costa sits down with Bruno Figueiredo, founder of UXLx and one of the longest-standing voices in UX in Europe, to explore how artificial intelligence is reshaping the design field.

From early web design to today’s AI-powered tools, Bruno brings a long-term perspective on technological shifts and explains why this moment feels fundamentally different. As AI accelerates tasks like coding, research synthesis and interface generation, the role of designers is starting to evolve in unexpected ways.

But while AI can generate outputs faster than ever and not all parts of design are equally solvable.

When we move from code to creativity, from execution to judgment and from data to human behavior, the limitations of AI become more visible.

This episode explores what AI can and cannot do in UX today and what designers, researchers and product teams need to understand to work with these systems effectively.

We discuss:

• Why AI may be the biggest shift UX has ever experienced
• Why code is easier for AI than creative design
• The limits and risks of synthetic users and automated research
• Why accessibility still depends on strong design foundations
• The growing problem of AI transparency and training data
• How AI is reshaping UX roles: specialists vs generalists
• Why AI should be treated as a collaborator, not a replacement
• What junior designers should focus on to stay relevant

//Human x Intelligent explores how humans and AI design, build and collaborate in intelligent systems//

Transcript

SPEAKER_02 0:00

So right now it feels like AI is redesigning the entire field of UX overnight. But today's guest has been in the internet since itself emerged, evolved, and even matured. Bruno Figueredo has been designing digital experience since their early web and is the founder of UXLC, one of the longest running UX conferences in Europe. So let's start with the big the big question. Is AI actually the biggest shift UX has ever faced, or have we seen this moment or a moment like this before?

SPEAKER_00 0:31

I think we've seen a shift uh a few years back when we entered a higher degree of specialization, but right now I think it's the biggest shift I ever witnessed.

SPEAKER_02 0:41

Okay. And what I find interesting about this moment is that many designers feel like everything is changing overnight, like I said in the introduction. But from your perspective, having lived through several different uh technology transitions, when you look at this current moment with AI entering design tools and workflows, does it feel like a true paradigm shift or does it resemble previous technology technological waves you've seen before? Because one of the things that seems to be driving this excitement right now is how fast artificial intelligence can generate things, especially code.

SPEAKER_00 1:18

Yes, actually um I come from an architectural background and uh I witnessed a tectonic shift while I was in college because when I started everyone drew by hand and then the cuff really kicked in. So it was a very labor-intensive job previously, and then with the with the introduction of uh new software and new ways of working, not only it sped up our work, but also it m it made it possible to design buildings that we couldn't have designed before. So I think we're witnessing kind of the same thing where our work will be sped up significantly, but also will free us to design and to evolve into more complex designs.

SPEAKER_02 2:00

Okay. And there's a lot of enthusiasm when it comes to AI-generated code or AI generated design and this shift of making everything fast, fast, fast. And one of our previous conversations, you mentioned something that I found very interesting and I wanted to circle back during this talk, but it's specifically that oh, actually that this is might actually be the easiest part of AI to solve, like the coding. And I would like to know from you why do you think that code is easier for AI systems than other parts of the design process?

SPEAKER_00 2:32

I think code is a little easier because code um is kind of akin to mathematics. So there's a truth to it in a way that if you want to accomplish something, there aren't a lot of different answers to get there, particularly if you're using a specific framework. But in design it it's different because there's a lot of ways to address a solution. And AI is built on is fed from solutions that already exist. So there isn't a lot of ways for AI to be truly creative. So I think it's a little different in that regard.

SPEAKER_02 3:08

Okay. And also that raises a lot of interesting contrast because when we move away from code and to the creative side of design, things suddenly seem much more complex, like you were saying. And when we move from code to visual design and creative work, things seem to move and move in a direction that can be more broad than we were thinking. But why do you think the creative side of design is so much harder for AI systems to handle well on contrary to uh the coding part?

SPEAKER_00 3:39

Okay, so there there are certain things that AI can do very fast and very well, and that is very simple challenges, like um, if I w if I opened up a new pizzeria and I want to design an app for people to order pieces for me, the AI will tap into existing apps that already exist. And for that specific purpose, if you're happy with uh with the visual design that uh that the AI produces, if you're not very demanding, then you're absolutely fine. But uh for a designer to maybe conduct and get exactly what it he had in its mind but produced by the AI, I think it's harder to nudge it in the AI into what exactly it has in its in its mind.

SPEAKER_02 4:21

Yeah, taking from there and the mind of designers, another area where I see artificial intelligence moving very fast, and we also talked about this previously, is research. We are starting to see tools offering things like synthetic users, like the synthetic user company, automated interviews, AI-driven usability testing. And I believe that in the surface it sounds very incredible and even powerful, but do you think that artificial intelligence can meaningfully replace parts of the UX research, or are there certain aspects of human behavior that simply cannot be simulated?

SPEAKER_00 4:59

Well, I think that whenever you design something, it's very unique. It's unique to the target audience, it's unique because no one has produced artifacts like that before. And since AI is built on uh data that was fed previously, it's hard for the AI to have a very good glimpse of how what could work in that regard. So, in terms of research, I think that um I helps a lot in synthesizing uh uh research data, it helps a lot in desk research, it helps moderating some uh uh usability tests. I've seen AI AI tools that um can also interpret facial expressions of the of the participants. If I think they are as good as a moderator, not yet, but maybe good enough. But regarding synthetic uh synthetic users, which means that you're extrapolating data from bodies of knowledge that you don't don't know because we we never know what's being fed to the to the it makes me a little worry because um uh uh I don't know what was fed to it, and usually the the good research data is locked in uh inside companies and inside um uh research companies, and that data is never uh publicly available, so it couldn't be fed to the AI unless they bought it, I don't know. But that that makes me worry a little bit. How can it generate good data if that data is not readily available?

SPEAKER_02 6:41

Yeah. And also if companies start relying heavily on these tools, like the synthetic tools, what are the risks that you think appear when organizations are relying too much on synthetic users or automated research insights and so on? I'm very curious because yeah, I saw that too. Yeah, go ahead. Go ahead.

SPEAKER_00 7:00

I think unfortunately the companies rely a lot on either on very dim research efforts or bad research data. For instance, a lot of companies rely on NPS, which is a very narrow tool, but they extrapolate it into all kinds of different levels within organizations. And uh and the the little the little research they do sometimes is also uh a very small effort in trying to analyze things. So and I think it's very seducing for companies to look at synthetic research and think, oh, I don't have to recruit users, I can get answers instantly. And the danger here is to rely on those tools and on that data and build things up upon them, which weren't exactly validated with anyone. So they they might be building things that are not exactly they don't fit the the market that they're looking for. There's that risk. But we're very in the in the very beginning of using these tools, so we don't know what the actual output is. I think we'll see that maybe later in this year.

SPEAKER_02 8:09

Where do you see the value of uh research in companies, even though we are having this huge boom with tools that can do a lot of the parts of research? Where do you see the value of a researcher, a UX researcher on a company organization?

SPEAKER_00 8:25

Well, I I I see a lot of value in doing research. I think companies sometimes are afraid of that research because sometimes the research is spearheaded by someone in the company that wants to somehow validate their preconceptions. And they push the company towards a certain direction. And if the research denies that direction, they they're put in the in a tough spot. I've seen the people speak of uh projects that happen in inside companies where the research turned a very different result from what the manager was looking for, and the manager just said, okay, forget the research, we're going to do what I wanted anyway. So uh I think there's a still a very big divide in the knowledge of what research can bring related to sea levels and uh and people in management in organizations that makes research be less valuable than the than it could be.

SPEAKER_02 9:26

Yeah, that's that also bears the question if the product they're building is more toward a long-term success, scalability, etc., or it's based on something egotistic, because most all of us have egos, right? It's a common human trait, let's say it's like this. One's have more, others have less, and others are more aware of their circumstances and what they are trying to build. That's also related to that. How how would you say that a researcher should handle that situation? Because how can they show their value right now if they were in that kind of situation?

SPEAKER_00 10:02

Well, I think the the the main thing is try to learn how to speak with upper management because researchers are very immersed in their medium, they speak their own lingo, and sometimes when they try to speak with upper management, they use that very same lingo, and upper management is like blank stare, they don't understand what they're saying. So I I think that we need to try to find a common ground with them. So maybe don't throw like the full card at them, maybe spoon fed them a little bit of information every now and then to saying, okay, we did this research uh and the main result was this and this, and we're very excited about it. So don't throw them like a full 130-page report. So give them one pager, show them one slide, and slowly ease them into uh the benefits of conducting research.

SPEAKER_02 10:55

And another interesting promise of artificial intelligence is for sure accessibility. And I know you have a lot in your mind about this, but some tools claim they can automatically fix accessibility issues or detect them early in the design process. Do you think that artificial intelligence can genuinely improve accessibility, or is or does it mostly work when good design foundations already exist?

SPEAKER_00 11:22

Well, I think accessibility needs good design foundations, and I also think that AI can fix some of the technical issues primarily related with code. Maybe not perfectly yet, because uh AI needs to fed on something, and there aren't a lot of good examples yet. For instance, I've looked into a website today that supposedly has a good accessibility score in automated tools, but then it was riddled with ARIA roles that are completely unnecessary. So I think AI can get there, but I don't think the developers that um that feed that data into the AI tools still understand how to do that properly to teach the AI tools to do it better. Where I see most promise is uh using AI in uh physical environments and um to help people with disabilities overcome navigation into physical spaces and other things that would be impossible a few years back. I think there's a lot of promise in that, and there's a lot of promise in, for instance, generating audio, translated audio in the same tone of voice as the original actor or doing closed captions. I think closed captions are getting very, very good. In English, in other languages, I don't think they're quite there yet. For instance, uh we we do a lot of research and we use a lot of tools to try to transcribe videos, and I would say that current tools are uh 70% there. Our language is uh in Portugal is very broken with a lot of vowels, so it's complicated for a to understand, yeah. Yeah, but I think we'll get there.

SPEAKER_02 13:08

Do you think that the reason that we are still so far away from having artificial intelligent accessibility inaccessibility is because uh it was never seen as something important. Also, related to that, or important no, or at more or as important as the other components, the other variants in a in a design process or in design uh visual design. And do you think that with the new UA Act within the accessibility sphere is going to help that because it's pushing that they need to do it or not?

SPEAKER_00 13:42

Well, I think the European Accessibility Act definitely pushes companies towards the the right direction. I think that uh until now the developers were very much focused on uh trying to translate the designs into code as much as possible to be pixel perfect, but they weren't very worried about the the the right semantic way of doing it. For accessibility to work, it needs to be built semantically. So I think they don't have a lot of knowledge towards that. I think they're slowly gaining that knowledge, those those teams pushed by the the European Accessibility Act. So my understanding is that in a few months, maybe in the beginning of next year, the developers will be at a different stage and may be able to produce better code, and in turn, that code fed to AI tools will make those AI tools also better.

SPEAKER_02 14:36

And in the end, it also helps with the royal of the business because if more people ha feel comfortable using their products, it's better. I went to a talk a few months ago related to accessibility, and I thought it was very interesting the number of it was a lot of a lot of the percentage of people with disabilities that want to use a tool don't use it because they cannot use it. There's one specific a friend of mine cannot see white. She sees uh rainbows instead of white, so it gives her a lot of headaches. And that's something that people don't see in the days of days, but there's a lot of per there's a lot of people like that. So I find it very intriguing that we haven't seen this as a business opportunity. So how would you present this to uh our accessibility as important for the business role to a C level or to a manager?

SPEAKER_00 15:28

Well, I think it's common knowledge that disabilities affect about 10 per 10 to 15% of the population. And it seems like a good chunk that you shouldn't ignore. But I think that when it comes to actually translate that into gains for a company, I think it's hard for companies to understand how much they could they could gain. I think also because there aren't a lot of good examples. If some big company spearheaded an effort into that and maybe they pushed out um uh a press release saying, Oh, we invested in accessibility and we got an increase in revenue of five to ten percent, I think others would would listen. And I'm just waiting for that to happen because eventually it will happen with this focus on accessibility. So I hope it happens as soon as possible for the benefit of all of us.

SPEAKER_02 16:18

Yeah, 100%. And also that brings us to something that doesn't get talked about enough, in my opinion, when we discuss artificial intelligence tools, because most AI tools provide very little transparency about where their knowledge comes from, what data trained them, are they selling to third parties, how reliable their outputs actually are. And how do for you, how big of a problem do you think this lack of transparency is for designers relying on these tools?

SPEAKER_00 16:50

Well, I think those tools don't want to advertise a lot of uh what was the data that fed them because some of those actions were illicit in some way. Because I I know I have friends who have books that are referenced in AI tools and they never authorized them to do so. So I think to prove their worth, a lot of those tools made some illicit shortcuts, and that's why they don't want to advertise them yet. But I think it's very important to be transparent uh about um uh about what was fed. And also I think there's uh an issue with that. A lot of information in our world is not exactly documented in books. Some of them are in our practice and some of them are in our head. Don Norman in the book uh uh Design of Everyday Things, he mentions uh two concepts, design in the world and design in the head. And what what he means is design in the world is what's documented, things that you can reference, and designing the head are things that you learn and aren't documented anywhere else. So uh for AI to be a really great tool, it would be uh the they would have to tap into the knowledge that's inside a lot of heads that uh never came out of it, uh, or are uh somehow dispersed into uh documents that they will never have access to. So a lot of human knowledge is not documented in a way that could be fed into the AI.

SPEAKER_02 18:22

And there's still a lot of documents that are just on paper. So just a parenthesis.

SPEAKER_00 18:26

So you're right, on paper and never digitized.

SPEAKER_02 18:30

Never. And I don't think they will. I have a lot of notebooks in my in my room, and none of them will ever see the any type of digital sphere. But I think because of these tools, or because if these tools become deeply embedded in the design workflow, they might also start changing the structure of the profession itself. And I wanted to talk with you over uh about this because you have over or over two decades of experience, and UX has become increasingly specialized. Because I remember when I first started doing UX by design, product design, it was called web design. It doesn't matter if it was a tool, for me specifically, it was presented as web design, and I was doing applications for mobile. And a lot of things have changed. And for me, this was like less than 10 years ago that was like this, and now we have this boom. Like uh even throughout the years, we had researchers, interaction designers, accessibility specialists, service designers, that most people don't even know what service designers is yet. And service designers one of the most important things in the whole process because it creates the whole hypo system. But AI tools seem to allow one person to cover many of these areas. Do you think we are moving into a more instead of a more special ID UX roles, if we are going back towards more generalists when we used to call the T-shaped or we used to call the Swiss knife?

SPEAKER_00 19:58

Well, I think so, because AI will definitely free us from those uh tedious uh tasks. And that means that we'll have more time to explore other ways of working with the tools. So let's say um for for instance, um I do some guidance for some of our clients. And I was uh correcting a designer the other day. She was presenting the work and I was guiding her into making some adjustments. And doing those adjustments takes such a long time that after we talked, she had to go over it maybe for one day or a couple of days, nudging pixels left and right. And it seems like something that AI, if AI were a part of the conversation I had with her, something that AI could automatically fix and then she would be free to do some other thing, maybe do research for the next project or do something else. So as we free people from those tedious tasks, the the breadth of work shrinks a little bit, and that enables us to concentrate on on the things that we're more prone to do. It doesn't mean that all the designers need to be unicorns, but maybe they will be a little less specialized. And if if you're a designer and you like to do research, maybe now you you'll have some some more time to be involved in it.

SPEAKER_02 21:23

Yeah. That's actually made me some curious, and I just wanted to give a note of what you said, like artificial intelligence as a co-work co-worker in a meeting, helping with solving the problems that a senior designer or experienced designer saw and pointed out. And the only the job of the person that is a bit less senior is to make sure that the artificial intelligence, the co worker, is doing its job while starting another job that is the next important thing because the other one is already being solved with the foundation created by the designer. Was just um a thought that I had out of what you said. But let me ask you a slightly provocative question. There are two possible futures people often imagine, and we have been talking and talking and hearing this, I'm sure you've heard of it. But one is AI designing everything automatically. This is the first world. The second world is the that AI removes the repetitive work so humans can focus more on creativity and problem solving. Which of these futures you would choose? And if none of them, what future comes to your mind?

SPEAKER_00 22:33

Well, definitely the second. I think the the the first one might be a pipe dream of some um of some corporate folks that uh would love to be able to, I don't know, fire all the designers and put uh put a team of AIs to do everything. So I I don't think we should um uh you to to do all the work to substitute us. I think we should use AI to work with us. So it's uh it's a slightly faster call that you have on hand. I think right now we're at a level that AI could be maybe an intern that uh that helps you. So it has the the the the same knowledge, uh the same technical knowledge as you, but not as much experience, and you can manage it to do some of the work with your help.

SPEAKER_02 23:22

Perfect. And we've talked about the shift of UI in these past years and how big it is and how big it can get. We've talked about a lot of tools, we talked about code, we talked about the creativity, we talked about accessibility, we've talked about transparency, specialization versus generalists. We've talked a lot about a lot of things. But to close this conversation, I'd like to bring it back to the people entering the field today. Because this has been very scary for them. And as you know, I'm I'm a co-leader for Ladies at UX in Lisbon, and a lot of the ladies that are starting or transitioning to UX are very worried because everyone is saying that we don't need juniors anymore. Internship, why would you do that? We need seniors, seniors, seniors. If a younger designer starting the career right now asked you, how do I remain relevant in a world where artificial intelligence can generate interfaces, code, and research insights, what would you tell them?

SPEAKER_00 24:23

I would tell them that they should look into try to use AI as a tool in their belt, but not as a tool that does the work for them. So don't expect to just tell them, okay, the please design 20 screens for the for this app top to bottom, but instead work with them. So to try to come up with uh with a creative concept and then try to guide AI into the several stages to design the the artifacts for them. So I think that's the key. So like uh if you're doing screen flows, uh to try to use AI to do the screen flows for you. If you're doing wireframes, try to use AI to do the wireframes, but all always with your agency, with your direction. I think that's the most valuable skill you can get right now.

SPEAKER_02 25:15

Thank you so much for being here with us today, for sharing all your knowledge. I'm sure people will have more questions and will want to reach out to you. So if everyone, if anyone that is listening would like to reach out to Brun Figueired, there will be in the description the LinkedIn and a lot of other information. But again, thank you, Bruno. Thank everyone that is listening.

SPEAKER_00 25:36

Oh, it was a pleasure, and feel free to reach out to me on LinkedIn.