Dan (00:00)
Welcome to AI and Design, where we explore how artificial intelligence is reshaping the world of design. I'm Dan Saffer

Nik (00:07)
and I'm Nick Martelaro, and we're faculty at Carnegie Mellon's Human-Computer Interaction Institute. Each week, we break down the latest AI developments, dive deep into topics that matter to designers, and talk with fascinating guests who are right at the intersection of these fields.

Dan (00:19)
Whether you're a designer working with AI or an AI practitioner interested in design, we're glad you're here. On today's episode, we'll be discussing Figma releases their state of the designer report for 2026.

Nik (00:34)
The New York Times notices that AI is personalizing the internet, but you have no say in it.

Dan (00:39)
An article by Jenny Wanger on how AI breaks flow, fragments attention, and quietly changes how product teams think.

Nik (00:47)
And finally, a new Harvard Business Report that says AI doesn't reduce work, it intensifies it. But first, let's talk about Figma's State of the Designer in 2026 report. Dan, is the news good?

Dan (00:58)
I would say that the news is definitely mixed. There are some good news in here and some bad news. It's a good time to be a designer in the Middle East or APAC, apparently, but less great if you're in Europe or North America.

I thought the skills that are most in demand was an interesting thing. number one was visual design and polish. Number two was using AI in the design process. I was a little surprised that that wasn't number one. And number three was designing AI driven products. So AI right there in the top two.

I would also wonder whether that visual design polish is because that is the one thing that AI is not as great at these days. So I wonder whether that is, that's why

Nik (01:53)
Yeah, I could imagine that. having used these tools fairly extensively now, still the visual polish, the detail oriented aspects of things, like there's so much going on that AI still isn't there with. I think it's better when you're working with a structured design systems, but even then it misses things. icons are weird fonts are oddly spaced and so much of

our role as designers is in that high detail level of craft. And that's something that the report really talks about. a lot of designers are really leaning into the fact of their differentiator being their high level of craft and that they're really pushing to have craft lead the story and lead the narrative within their organization. Also, I think that that aligns with how

even people who are doing a lot of AI-oriented design work are thinking about things. We were talking about Jenny Wen at Anthropic, and in her talk, she talks about how even they're focusing a lot on craft. What makes an experience delightful? What pushes the experience to be highly polished so that it really is an excellent experience for the user?

Dan (03:02)
And the Figma state of the designer did ask what is craft? Number one, visual polish. Number two, thoughtful problem solving. And number three, clear intuitive UX. I personally was pretty happy about this. I'm glad to see thoughtful problem solving and intuitive UX way up there.

I think that those are the things that are always going to be part of the toolkit, even as the AIs get better and better and better at visual polish.

Nik (03:33)
One of the things here too is that in the report they say that AI was improving how designers worked. 91 % of designers said AI tools improve their designs. And I could imagine that even if it's not doing as well in sort of that high level of visual polish, there are a lot of ways in where AI might be offloading work for you such that you can focus more on that really thoughtful problem solving or really figuring out what makes something intuitive.

Dan (03:58)
related to that a thing that I noticed was that designers who aren't increasing their AI usage were twice as likely to say that their job is getting worse and Do we think this is a case of?

AI actually making things better, like you were just saying, or does it reflect some kind of selection bias where more optimistic designers are simply more willing to experiment?

Nik (04:22)
Yeah, that's an interesting point on this split and selection bias. The survey itself was pretty broad, 906 digital designers all over the world and actually pretty evenly split all over the world. And yeah, I do wonder if maybe anyone who is sort of excited about AI, maybe they were more, one, willing to answer these questions and two,

that they're just going to be more positive about things because it's new, it's exciting. It's also possible that they're working in fields that are willing to accept this, they're willing to try things. Whereas if you're not using this, it's possible that your organization doesn't allow you to do this and you're looking out and seeing, well, the world is changing around us, but like we're staying static. Yeah, I don't know. There could be all kinds of reasons for why that split exists.

Dan (05:04)
Mmm, mm-hmm.

it could be an age thing too, where it's like, this is challenging. And it's this thing that I have to learn, at a period where I'm older or I just don't have the time to learn it. That's certainly something too. You may not even be older, but you may have a young child at home. And it's like, God, another thing I have to learn in order to keep my job

Nik (05:25)
one of the takeaways from the report is that folks who are trying AI, it's not destroying anything within their process. It's not taking away their craft. They seem to actually be doing what they believe is better design work. And it seems that maybe what

The recommendation from the report is, like, you should be experimenting. You should be trying things out. and I, I kind of agree with that. I kind of think that this is an opportunity. It's still early days really, even though it feels like we've been in this for now many years. I think that building up what they kind of call this culture of AI experimentation is probably worth it. and it's probably worth organizations to at least be setting aside a little bit of time just to see. Cause

The cool thing about it is that you don't have to think about AI as like, this has to replace all of my design processes, has to change everything. you can just put it in at one point and potentially that improves one aspect of your design process or your team's process, and that helps maybe accelerate you or improve the work you're doing. that can be a really good step to take.

what I'm seeing here, is that there's actually a lot more positivity around AI than I actually had originally anticipated. I thought when the report came out, before I opened it, I thought it was going to be, AI is really messing things up. It's really causing a damper on things. But it was really interesting to see that, at least of the respondents in this study, people seem to overall be saying that AI is improving the work that they're doing in some way or another.

And I think it's worth maybe taking this. And if you need

Dan (06:54)
I will say that some of the bias here these are definitely people that are using Figma and these are definitely people that are probably very pro tool and pro Figma and Figma has a vested interest in your using their tool and experimenting with it. So I would definitely take some of their findings in this with

A little bit of grain of salt. Not that we don't love Figma, but they are not an unbiased neutral party either.

Nik (07:23)
read the methodology section of any study like this. who are the respondents? Where are they from? I'll say that it would be nice. I think the report was a little thin. I mean, I really like the representation across the world. That's super cool. But, you I don't totally know who all of the respondents are.

And you know sort of where they're coming from what industries and things like that and it would be interesting from a data perspective how would how would things break down Are there breakdowns with age are there breakdowns with industry? You know, are there differences and that would be interesting to understand

Dan (07:55)
The next story we're going to talk about is from the New York Times and the title of the story was AI is giving you a personalized internet that you have no say in it.

The point of this story is that tech companies have been making a personalized internet just for you, but they are pushing it at you and there is no off switch. There is no opt out. You are stuck with this layer of AI, whether you want it or not. And for most people,

Finding that opt out switch or turn things off is gonna be beyond them because as we know as product designers, no one ever changes the defaults. only power users will go in and really change things.

Nik (08:39)
I generally agree with what the author is trying to get at, which is that a lot of AI features are being pushed on us without an ability to opt out and that they are starting to change the way in which we experience the internet. That being said, they've been changing the way we experience the internet since we've been making tools to experience it. the other thing that I didn't totally get from the article was actually this idea of personalization.

It didn't seem like there was as much engagement. Well, is this good? Is it bad? Are they doing it in a way that is good and positive? I mean, I might argue the internet is a huge place. Having something personalized for me where it really understands me, if I have an AI agent that understands me, example, could make it a much better experience for me.

Dan (09:22)
I've worked at a couple of services and one of the things that was very hard for people to understand was that if you weren't paying for something, you didn't own it and the service could change it at any given time. And I think people do feel a connection with

services, whether that's social media, or whether it's Google search. And so all of a sudden seeing a major change can be very disruptive and very anger filling, particularly if you're not feeling that you're getting a lot of value in it. And I think that's the real problem is that feeling that AI is being stuffed at us and not feeling like it's valuable.

That's the hard part. I do remember back when Google started doing this, was it last year? Was it early last year? Was it two years ago? It hasn't been that long, but how many people were like really hating on it, really not liking it.

So it's either that they're like actually now finding value in it or they've just come to accept it as another piece of and shitification of the web. I can't I can't really tell that but ⁓ but I know for myself that as the results have gotten better and more personalized at the top of Google search, I don't mind them as much

Nik (10:23)
Now, if I do not agree with you, or they just want to take back and have another piece of education, I can't really tell that, I'm really something that the results have gotten better and better. First one is that the bottom line like this as

Dan (10:47)
I think that's really the key to all of this is if the AI is valuable and it's personalized and it's saving you time by being efficient, then

Nik (10:48)
the key to all of this in your environment, the AI is not any personalized and it's safe in your time by being efficient,

then it's we will come to an end. It's when it's in ways, doing things that give one or two

Dan (10:58)
It's great. We don't complain about it. It's when it's gets in the way. It's doing things that you don't want it to do for you

that it is just another layer that you have to pass through to get to the thing that you really want to do. That's when it becomes really annoying.

Nik (11:16)
Yeah, one of the interesting points in this article that is much deeper down in it is how they talk about chatbots as a way to get more information from users in the wake of a lot of other data protection things having swept over the internet such that you couldn't get as much information.

You can't track as much data from people. You can't store it because of new privacy, rules and laws. But now people are much more willing to give their information to a chat bot. And, from the perspective of design, this is amazing information about our users is what we need to create better products. In the article.

They specifically, though, call out that this is more about advertising. And this is always a tension, right? I think this is always a tension in the same data that can be useful for a designer, effectively as a form of user research. Basically, can also be effective for a marketer to potentially sell you something. Now, maybe it's something you wanted, and maybe it's something you didn't want.

Dan (12:11)
I do like the metaphor that one of the people quoted in the article says. says, AI optimization is like plastic surgery. You notice the bad stuff, but you don't notice the ones that are like really good

But yeah, I do like your, contrast between marketing and design and getting information. This is something that Shane Johnson talked about in our episode last week was that he was afraid like that AI wasn't going to be as good as he was about getting people to talk. And it's better because people will just tell AI anything. No, I don't recommend doing that.

But people are just more open and will say things to a chatbot that they're not going to say to a human being.

here's a funny anecdote. was actually, I did a Google AI search trying to look for unusual Valentine's Day poems over the weekend. And

Nik (13:11)
called. That's what makes you want my attention on the internet. So I'm going to about this. I'm to talk about the book I'm writing. I'm going to talk about AR. I'm to talk about one of my hobbies. And I'm going to about my day. I want you to know that I'm staring at

Dan (13:11)
that were based on my personal interests. So had a bunch of things about some of the stuff on the book I'm writing, some of the stuff about AI, some of the stuff about one of my hobbies and wrote Valentine's Day poems for those. And I thought, well, this

is a funny kind of screw up. And I'm like, no, I want other people's poems, not you to write me them.

Nik (13:35)
That's funny. Although it does speak to an interesting development that we might see in the future of design. The fact that things can be designed in real time for you.

based potentially on your needs, wants, desires, interests, right? The fact that it can design or I guess in a way it's like designing a poem. I it's writing a poem for you, but it's creating something with a goal. And I actually think that that could be a really interesting future. We've talked about this in my group here, where can you have some form of

AI and or human AI team that really allows for the creation of personalized products for people that are designed for them. I've used it in the example of a kitchen table, Beautiful handcrafted kitchen tables are lovely, but very expensive. And most people can't afford this.

I wonder, okay, well, why not? Why couldn't we actually have something that understands you, understands how you use your kitchen, understands your budget, understands your aesthetic style and preferences, and then crafts and creates something for you? It designs something for you. And this is different than mass customization and you going in and sort of configuring a table or something like that. Like it actually sits down and design something for you, which I think is a really lovely experience,

think about the last time someone actually worked with you, like you worked with another designer for them to design something for you is a great experience in my opinion. configuring a bunch of stuff yourself, like that's actually tedious. You probably only like to do that with like the one thing in your life. Like I like configuring, picking my shoes and customizing all the colors, but that's for like, one thing, but actually someone sitting down and saying, well, Hey,

What do you actually want to do with your shoes? Where do you want to go? I mean, it's funny because there's an example in the article about a chat bot understanding that you like running in the winter and then ads being created that basically push products for you that say this is going to be a good shoe in cold weather. But you could also look at that from the design perspective of like, wait, this is information to get back to the team at shoe designers to say.

we've got a bunch of people who like running in the snow and actually they're complaining about the durability of their shoes. Maybe we should figure out how to design something for them that's gonna work. it's two sides of a coin, but I think right now a lot of things are focused on marketing because marketing is where the money is. But for us as designers, and I hopefully maybe for our listeners, like I do think it's worth.

thinking a little bit about like, could this mean for the future of design practice? When in theory, you could have AI systems in the same way that it's basically creating custom poetry based on your needs. What if it starts creating products based on your needs?

Dan (16:11)
configuring things is work. And I'm always telling my students, don't make your users do work, which is why, no one, no one changes their defaults, getting back to one of the core things of this article. but personalization or something that is made just for you, that, that definitely feels bespoke and special and

all those good things.

Nik (16:33)
The other thing too here is that customizing things is work, work is hard, but actually having a conversation with someone that can be meaningful, it can be positive. I think back actually to my thesis work. So I guess for listeners to know a little bit and probably so you understand my biases and what we're talking about here, my thesis was all on thinking about how we might use

conversational agents out in the world as a way of understanding user needs, being able to observe. Now, when I did it, it was actually all about remote user research, but under the guise of an AI voice assistant that was there, but it was really designers all on the back end. And one of the things that we learned when I was doing stuff was that

people actually really liked having that conversation. And I think this is something we're seeing, right? A well-crafted conversation with an AI agent is, it's fun, it's enjoyable.

I think that you can have a great experience while creating a product for someone that ultimately they will want to potentially buy.

Dan (17:33)
All right. Our next story is by Jenny Wanger called waiting is the new interruption. How AI breaks flow fragments attention and quietly changes how product teams think. And the premise here is that when AI is taking 10 seconds to respond or 30 seconds or two minutes.

that kind of rhythm. It just invites distraction. It invites you to jump away and do something else and invite you to lose your train of thought. It invites you to just go do something else or context switch and then you come back and things are and you're and it just takes you a while to get back into the rhythm of it. So it just keeps breaking.

flow after flow. know I've definitely experienced this Nick I'm sure you have.

Nik (18:30)
Yeah. I experienced this all the time because I'm often using systems that are in their thinking mode. And so the responses are taking a while to get back. When I'm working with code assistants, some things are minutes, a job could be running for, quite a significant amount of time.

And I definitely feel what Jenny's talking about is that I'm now not in a flow state. And actually I'm not totally sure what to do because depending on how fast it is, and I also don't know how long it's going to take, I kind of am like in this weird of like, well, do I check some text messages? Do I open up my email?

this is not an experience which feels ideal for me. Unlike other kinds of work where I'm really locked in and really focused and in that sort of productive flow state.

Dan (19:12)
Jacob Nielsen wrote an article a couple of months ago called Slow AI. He's talking about much longer timeframes, but they both are queuing into the same problem. And the solution that we have been handed right now is just wait patiently user, it's coming.

And that's just unrealistic given how human beings are wired and that context switching has these real cognitive costs. And the current crop of AI tools just hasn't really addressed this waiting problem at all. And that threshold that 10 seconds before you're checked out.

And even one second before you feel that interruption is really important. But you do lose that thread. And she calls it lost focus. Nielsen calls it cognitive discontinuity. And I think what's interesting is that they kind of are coming at it from

different angles. seems like she really wants to treat this as a personal challenge. And I'm guessing that Jacob Nielsen would see that as a bandaid over the design failure of the AI tool itself. If the tool is forcing people to develop these

Personal coping strategies for its latency then the tool is broken now, I don't know how to fix that given the Time it takes to do these kinds of deep research tasks or agentic task that do take hours upon time this seems like a super ripe area for

some academic research.

Nik (20:54)
this is where there is going to be hopefully some really interesting interaction design work coming out. How do you actually design the experience that people have when working with long running agents, longer running tasks, things where you don't know how long it's going to take to get back a response? Because that's where right now

the experience is kind of breaking. And most people are just putting up with it. Actually in the article, there's a really nice figure here of a bunch of strategies that you could employ. And I liked a lot of them. One of them, for example, was about micro-switching, where it's like, okay, I'm gonna take a break and I'm gonna go do another test. That's kind of probably what people's default is. But there were some other ones I hadn't really thought of.

Parallelization was one, rotating between projects, like actually having multiple projects. Now, I'll say there's some risk there because context switching can actually just be cognitively hard and it can be draining on you. ⁓

Dan (21:46)
Right, that cognitive

discontinuity.

Nik (21:48)
Right. But then there were other ones here that like the two that I think I liked the most that I might employ today are the embodied reset and mindful waiting. So the embodied reset was literally like stand, stretch, move, get in touch with your body, do something else, which I thought was kind of cool. And I don't know why I hadn't really thought to do that, but actually it's like a good reason to get up from the chair and move. And then the mindful waiting one was interesting to say,

you know, okay, what am I thinking about? What are we doing? And let me just like, try to focus on it almost in more of like, let me meditate on what it is we're doing. And then when I get the response, right, maybe my mind is a little more prime, because that's the other thing here, right, is that when you start doing either parallelization or micro switching, again, you are now thinking about something else. And then you've got to bring yourself back into the context where maybe it makes sense to just calmly sit and wait. Now the problem here is

I don't do look weird if you don't have if you're you know in an open office like you're working but then you're just kind of sitting there and like

Dan (22:50)
Yeah, I'm zen meditating. right exactly

Nik (22:53)
I It's probably just weird though, because people might look at you and go like, well, you're not thinking the AI is thinking. It's no, no, no, we're thinking at the same time. I'm just thinking of something else.

Dan (23:04)
Yeah, mean, these all still feel like these are human solutions to a technical problem. I wish there was a better technical solution or a better design solution to make the waiting ⁓

Less, empty and maybe that is my bias being like, we got to be productive, productive, productive. rather than this slow down, think about it, so that you don't have to do context switching or going away and doing something else so that you need what, Nielsen calls the context, reboarding where you have to

get back into the flow. But this always gets us back to that, this like productive speed versus like shallowness or shallow thinking that feels productive where I wonder if when I come back, I'm because it's taken so long and because I've jumped away, I'm much more likely just to say,

This is good, go, rather than really think about it and absorb it, I'm much less likely to catch any mistakes, because the AI responses often sound very confident and complete. It makes these errors hard to catch, particularly when you come back to them.

And so how do we build in safeguards against this beyond just telling people, pay closer attention to this because we know that people won't and don't.

Nik (24:31)
Yeah, I think that's a great point. I guess now I'm getting excited by the fact that there's actually a really cool interaction design problem to solve here. And this is going to be really fun for some set of designers out there to start thinking about whether they're tool designers. we're still early. mean, right now, we have basically the sidebar chat chat bot model. That's what we have.

we're also just playing tennis. know, the ball goes back and forth, we pass it back and forth, but actually I think we need some different metaphors. And I think we can start thinking about different interfaces that would avoid the problems that we're talking about here while also hopefully making us maybe more mindful and more aware of what's going on, more thoughtful with what the AI is responding with, and then ultimately more productive with the work that we're doing.

Dan (25:14)
Mm And my UI for AI team last semester did a bunch of this work around context switching. So I advise everyone to go check out some of that solutions, how they kind of thought of like, how do you get back into a task and stuff like that. So there is work being done on this, but it's still pretty nascent.

Nik (25:37)
Well, the other thing, Dan, is this really connects well with our last story and a result that's come out, which is that AI is not reducing the amount of work. It's actually intensifying the work that we do.

Dan (25:49)
Was this at all surprising to you? For me, I'm like, yeah, duh. using these tools, I feel like I'm back in COVID times when there was very little blur between work and life because you were

just working a lot all the time. the article describes this loop of AI speeds up tasks, and then the expectations for speed rise, and then workers rely on more AI, and the scope of work expands and work intensifies. And

That's exactly how it feels like to me. And I'm sure how it feels like to a lot of people listening to this podcast, which is why we're seeing so many people getting AI fatigue and burning out on this stuff, because it will take up as much time as you let it. Like it's that old adage of work expands to fit the amount of time that you have. And this will take up as much time as you have entirely.

Nik (26:46)
Yeah, that's been my experience. I mean, you you asked was I surprised? I wasn't entirely surprised. And part of it is because in human automation interaction work, say in like physical spaces where, you know, machinery automation,

robotic automation type stuff where there's still a human that's involved in a work task. Oftentimes we do see work intensification. And so it's not surprising to see that you're seeing that work intensification here. I think one of the interesting things that this article points out and relates to the last story we talked about was this idea of more multitasking. And again, I think part of it could be because

the way in which we work with these tools right now is this very back and forth. We're not kind of working at the same time. We're sort of handing off work, waiting for a response. And so as those response times get longer, as jobs get longer, you're going to be trying to do more different things because you want to fill your time with something.

Dan (27:37)
The thing that the article makes pretty clear is that the researchers found that the workers, voluntarily expanded their workload. It was not because like their managers said, go do this. Although certainly we're seeing in the tech industry, lot of managers saying use AI or you're gone. But they were using it because AI made them, made feel doing things like more possible, more rewarding.

And the commenters said, hey, that AI makes them feel superhuman. They're capable of doing things that they couldn't before. That feeling can be genuinely addictive. But like all addictions, think, yeah, you can burn out on it. And I don't know whether that kind of

Constant productivity that constant work is a net positive for workers or for the organizations it creates I think new risks around quality and accountability and risk for like well we could easily do this job that currently three people are doing with just one

Because look how productive everyone is getting which leads to absolute burnout.

Nik (28:43)
I definitely know now that

when it comes to creating prototypes with a code agent, like I definitely can do a lot more. It's amazingly empowering. And I now also feel sometimes where I like, wow, I can do this. It's so easy to do this. I have to do more. I have to make more iterations. I have to do more prototyping. Whereas before like actually the act of say, going into a tool and laying out something by hand would just take me a while.

But that work can actually be somewhat meditative. It could be a way for me to also slowly engage with my work and slowly form my thoughts. Like one of the other things here is that when you're working with AI systems, it all comes to you at once and you now have to respond to sort of all of it at once, Whether it's a response from a chat bot or whether it's, a new prototype, this actually may be more slow building and this more say slow reflective loop. I wonder if that's getting

kind of wonked out here and that may be part of one the addictiveness of that feeling of like wow I can do so much but also the thing that really leads to that intensification.

Dan (29:46)
there was something about doing things like laying out a flow that you know, you spend time doing it. And as you're doing it, you're like, Ooh, well, I thought I thought of something different or something better. Or here's this, other pathway I didn't consider. And it also gave me a break from doing some of that, really strategic thinking

Now there's like no rest for that because it's like, OK, I do do that heavy strategic thinking. And then maybe I'm pushing the button and it's going and I get three minutes and then here it is. Here's the here's the thing and then you had to go back and do more of that strategic thinking where before.

I might've had a couple hours to really be mulling over different things or just kind of doing hands work that is also like giving my unconscious brain time to process and think and do thinking while I'm making. And that is definitely getting lost. So it's hard to think about like what could start to bring back some of those

thinking tools or where the AI is making things that aren't finished, that are more unfinished and letting the user really finish it or really have to think about things before they finish it.

Nik (31:07)
one of the things that you're making me think of now is the way in which I work when I sketch by hand and how my thought and my actions basically are flowing at a speed that effectively my arm and my brain can kind of operate at and they're flowing in a continuous way, arguably the continuous stroke of the pen. But that's not really how a lot

of AI product experiences are today. Everything's really choppy, Back to kind of this idea to the other article, again, this weighting is that choppiness. And that's not really how I learned to work. And I'm not sure, like cognitively, if that's even a ⁓ great way of working. I'm sure there's probably some cognitive science research that shows that, and this is why you feel tired.

with this type of work, whereas when you have that continuous sort of flow. And so from the perspective of designing these types of tools, like if you are out there designing AI enhanced products, I wonder if you can start to explore what is sort of the human speed and the human smooth flow of work that you can employ into your tool as opposed to this really choppy, spiky, everything happens when it happens and then you wait.

Dan (32:17)
definitely some of the human AI collaboration research I've seen the HA I see a lot of that is around exactly what you're talking about where it is How do you

break up these tasks so that they aren't these like super long jump to the end kind of things, but it is how do you have these smaller moments that can maybe be faster and are part of that keep the human being in the flow versus these interruptions that keep pulling you out of the flow? Are there ways that we can start to make things?

faster and smoother. And maybe some of that is when more models move onto devices, we will get more of that rather than the constant call to the cloud that does create some of this latency

Nik (33:05)
Well, I'm excited for now that we've discussed this, I'm want to either try maybe looking at some of this stuff myself. But also for folks out there, If you're designing this kind of stuff, like I'm always excited to see all the new experimentation that people are doing with

changing up the interaction design for these types of tools.

Dan (33:25)
If you're designing stuff like this, reach out to us. Yeah, we'd love to have you on the show to kind of talk about some of the strategies that you're using to do this.

And I think that's our show for this week. Next week we'll be back with more stories around AI and design and we'll see you then. Thanks for listening.