Dan: Welcome to AI and Design where we explore how artificial intelligence is reshaping the world of design. I'm Dan Safer.

Nik: And I'm Nick Martillero, 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: Whether you're a designer working with AI or an AI practitioner interested in design, we're glad you're here. And this is a special episode where we talk about all the big themes of the first half of 2026. A lot has happened, and we thought this might be a good place to take a breath and reflect before the second half comes at us. Nick's been to Kai, I've been to Config. We've had 25 episodes and 10 special guests. And basically Think about what have we learned, especially as Nick and I are preparing our lesson plans for fall, and what are we teaching our master's students, our advanced students this fall? Something we need to figure out based on what we've learned in the first half of this year. But first, Nick, you and I have been playing with Fable all week. What have you made? What do you think?

Nik: I have been trying to leverage as much as I can of using Fable before they supposedly takeaway access on the nineteenth now. It was supposed to be the twelfth, but they extended it for us. So I guess we get to play around for another week. and I've been working on a couple things. I've been building some apps for myself, and then I've also been working on some of the apps for the little startup that I have. and it's been impressive. and in some ways, I don't know how different I feel about it in regards to its capabilities in producing output to Opus four point eight on high, but its ability to do long running tasks where I basically say Here is a whole product requirements document. Go work on it for forty-five minutes, an hour. that has really shifted how I've worked. on the show we've talked about this do something, chat and then wait. And it's interesting because that number of waiting has just been getting longer and longer. back when we first talked about it on the show, I think we were waiting for a minute. Maybe two minutes I was waiting for a response sometimes. I'm now waiting for forty-five minutes for something to come back. Now it's because I'm saying build the whole app for me and then bring me the full high fidelity prototype, but it's really been changing how I am working and how I'm thinking. and so now I'm doing this weird thing where I'm running multiple agents where we've heard about people talk about this. This is becoming very common. But I feel foggy about it all. it's weird. I've I had this feeling of okay, I it it's yeah, it's now my software developer, but I really don't know all the stuff that's going on in the background there. and that's been kinda weird. At the same time, some of the stuff that I've been getting out of it is great. it's really working well for me in regards to at least creating front ends. it's still doing really stupid stuff. I had one thing where it was when you highlight this as a link, it's just going to center the text because yeah, that's the style we should put. And I'm that doesn't make any why would you do this? And then I want to sit there and be you built this whole thing with an entire back end, a full AI connection, all this stuff, and you still can't do basic visual design. Dan, what have you been building?

Dan: I spent some time building a ambient bird listening app so that you can as you're walking around hear bird song through your airpods and it detects it and says, that's a cardinal, that's a whatever. Just super basic. But you, yeah, had the experience pushing the button and that first build took and it wasn't even that long, it took I don't know, ten minutes or something that. But still that was by far the longest time I've ever had to wait for something because yeah, I gave it a long or semi-long PRD. And it built everything from scratch and put it up and then it was all my time subsequently has been tweaking. there was that twenty percent of time to get it on its feet, and then just an enormous amount of time to get it right. And I'd forgotten about that. that's what that's what programming is, and that's what design is, and that's what all creativity probably is, getting that first piece up and it's a mess, and then you have to shape it and make it better, make it good, and so much of your time is spent doing that. And yeah, it was a really weird experience. Because at the beginning when we started this podcast, if you'll remember I did one of these to make a quick app to turn my newsletters into a daily podcast that I would listen to. And The app sucked. It never got good. I abandoned it after a while. I soon stopped using it just because it was doing terrible summaries that just didn't feel good and the app looked crap. And this time, no. I'm I'll put this on the app store. People can use it. They can play with it. it's good enough for that. we're talking four or five hours here. And amazing. And I have built web apps and stuff. I've never built Anything in Xcode and boom, done. it's astounding how fast and how much better it is than six months ago or maybe even six weeks ago. Just really incredible.

Nik: I think one of the things that I am recognizing is my ability to come up with an idea and create that first run prototype it is so fast for me to be able to do that. But also I have to sit there and, shovel in either my ideas or I have to get to the point of going back and forth to fix things. And I'm now I reached this point over the weekend where I was I have to sleep. I actually need to sleep. I can't go anymore. But if I don't keep putting things into the machine, it will not be building. And there's this weird feeling I have of but it could be building. And part of it could be that, I may not be the most advanced user. I've heard of stories from really professional software developers who have, runs that take six hours. Right. basically they've got it spawning dozens of agents and it goes for hours and hours and hours to build. And so you actually can set up a job like that and go to sleep and come back later. I'm not doing anything of that level yet. But yeah, it was a weird feeling. I had this feeling of I can create so much and yet I actually had this really reflective moment where I was I really need to sit down and almost slow myself down more and say, what do I really want to build? I think before when it was so hard for me to build, I did that maybe more naturally, whereas now I actually have to be much more explicit about it. in prioritizing what I think is important to actually create.

Dan: think there is definitely this addictive quality to it that it does feed that part of your brain where you're just well I could keep adding this. I could keep adding this. Cause then I was well I'm gonna add the watch to it. I wanna add some voice commands to it. I'm gonna and you just keep going and When am I done? when it when have I exhausted all the value here? And before anyone even sees it, before a user even touches it and gives any feedback to it aside from me and my pal Claude. who detects almost nothing, who's like, This is great, this is amazing. now luckily, I've been doing this a long time and it's pretty easy for me to pick up things that are janky and things that you don't need to see and all that stuff. But as I'm doing this, I'm like well if I can make this in, I did think I did four hours on Saturday and two hours this morning or something like that. I just Cranked it up in the two hours this morning we're definitely Well, what more can I add to this? What more can I be doing? Because like you, I woke up and was like, Well, I could be this thing's been sitting here idle, I could be burning tokens right now. And yeah, it's such a it's a weird feeling. that I don't particularly like. But I I don't like being a slave to the machine. I like feeling okay, this is good enough. It's done. Let's get it out there. I wanna see people using it or not using it, in which case I can discard it and be done with it.

Nik: Yeah, that feeling you're talking about of this, slave to the machine in a way, I guess I also felt similarly. that's what you're this feeling of your it doesn't feel great. whereas before I thought it would have this feels amazing, right? I can build whatever I want. And actually when I was using the older models and I was doing stuff in a very part by part manner, I did feel like that. I felt and I'll say it still does feel great to actually have an idea that I have in my head on the screen realized. that's is near nearing magical. But

Dan: Yeah, it's deeply empowering and that's the that's the flywheel that's driving all this, is you could type some words in and something appears on screen and it's amazing.

Nik: Yeah. But I think I don't know why it is with this. Maybe it's just because Fable pushed me to push it more. Maybe it is actually doing more. I do feel it is actually more capable than other models. I know there's lots of reports online of Fable's been it's not as good. they basically made it worse after it got pulled and then they brought it back. I'm not sure. I believe that actually. I feel it is pretty darn good. But yeah, I don't know. I actually this was a very reflective weekend for me. I felt I need to sit down and actually think a little more about how it is that I wanna design and how I wanna work now because I we've hit another sort of inflection point in the capability. and I think that I I need to consider what my new process is. But I already had a new process as of a few weeks ago, which was completely different from what it used to be a year ago. the fact that I'm changing my process again too is also a little I don't wanna say tiresome, but it is tiring.

Dan: Yeah. No, there is the work that you want or in the case of most of our listeners have to do and then there is the learning and scaffolding of all the stuff now around the work of building all the tools for the tool to do the thing. This is exhausting. we talked about this all the way back at episode nine when we talked about brain fry. they named it. I think it was Harvard Business Review, HBR, named it Brain Fry and we talked about this a long time ago and a long time ago. We talked about this three months ago. It was forever. but

Nik: Ha

Dan: Yeah, but brain fry is real as I was doing this, I'm well, what are we teaching students this fall? Because I know you teach AI Aug augmented designer, I teach advanced interaction design. what are they gonna need to know coming out of this semester at the end of it and how are we teaching it to them and what can we expect in six months time? if we're able to do this right now pretty fast, what will they be able to do given Two months of a semester with guidance

Nik: Yeah, I'm having to really think about what it is that I teach. and I basically have to change the syllabus almost I'll probably be changing it in real time. Typically I do, because while there's core aspects of hey, let's introduce you to AI systems and how you can work with them as a designer, the tools always change. one of the things that I think I'm gonna be really focusing on this semester is a lot more on this reflective practice and getting people to really think about what they're doing because you can run so, fast now, but you can also run fast without understanding why you're running so fast. I think actually people understand where they're running. They know they kinda know and they could be running as you've said on the show. I the way you talk about it and others, you could run really fast in the wrong direction. I'm actually okay with them running in the wrong direction and then getting a crit and then being yeah, that was not good. Cause then they can just, fire it up and two hours later they'll run in another direction. This is all very low cost, in class. That's the point. It's

Dan: And low risk, exactly.

Nik: Exactly right. That's the point in class, right? I want you to explore. I want you to run in the wrong direction and recognize that you did it so what that's And so but I'm asking students to really focus on that. yeah, I'm starting to sit down and think maybe what I need to do is I will work with these things, I will work fast and let them run and work for hours, and then I will step back and really try to reflect more. Whereas I think before it was a lot of reflection in action. now I think I need to step away and do reflection on action because I'm not sitting there moving at a pace where I can do something and reflect and do something. I sort of have to say, Okay, you go off and do something, I will wait, and maybe that's when I'll reflect on things, or I have to come back.

Dan: for me, I have a lot of big topics that I like to cover in the class. things that are foundational. Things like how to frame a problem, design systems, those kinds of things. And trying to figure out how to Thread AI through those because that will be some of the tools and medium that they're working in is my real challenge. Even as those things are changing, one of our big themes, this idea of the collapse of fidelity. now that AI has made high fidelity free. Do I bother teaching wireframes anymore? Is there feedback that we can't get from a polished artifact anymore or a polished prototype? And why should we still teach this? are these good thinking tools even after they die is actual deliverables or actual things that you are showing to clients or stakeholders. Are they things that just help people think through the problem?

Nik: what are the boundary objects that we use within our practice. And I would argue that a wireframe, a low fidelity mock can still be a good boundary object within a team or potentially with a client, except for the fact that now everyone knows you can create the high fidelity version. And you do get a lot out of it. Now, of course, there's all kinds of work, That shows, hey, if you show a high fidelity mock-up and someone goes well, I don't really the color, and you're don't worry about the color right now. I was mostly trying to get the layout and also let's think about the user flows. You're yeah, but I don't really this purple. Right, Dan, if you get a purple thing, right, you're just gonna sit there the whole time and be I do I we can't do it. But so

Dan: Right. Stewin' that. Why? Why would you ever put this color in here? I would have been I would have been terrible in those first Yahoo meetings, I guess. That's the only brand that has purple is this dominant color that I can think of. All the others rightfully went out of business. I mean even Yahoo is kinda going out of business, sort

Nik: Yeah. Roku. Yeah. and so I still think that these are probably useful as boundary objects. At the same time I there probably are a lot of instances where you also can get good feedback using high fidelity mockups. Or if you are working in a real product that exists, showing the new version as the high fidelity mock, just showing this is what the new product will be and getting the feedback is what you want because you aren't changing the colors. You're not changing a bunch of things about that. And I think that's one of the things to remember about a lot of the work that we do as designers when we work on a product that exists is that we're updating that product. We're not always creating new features. So to me, maybe when I'm creating something totally

Dan: Well, we're not always creating a new design system. It's the design system is given to us.

Nik: Exactly. And so in that way, I think maybe a lot of our bread and butter work, maybe it does just always live in high fidelity world now because it basically doesn't cost anything. And in some ways you can really see an experience in the full product context. And so i it's a good way to do this. But yeah, maybe when we're doing new design when we're doing really green field things, then we go back to these lower fidelity boundary objects as a way to communicate and to work through things. That being said If you actually are working primarily with designers who know how to not get hung up on the hi-fi stuff, can you work in the hi-fi? this is something, for example, I can do. I can turn off the don't worry about color, don't worry about this. And so I don't mind working in hi-fi, because actually a lot of what I want to do is test the interactivity. And this is something that we haven't had for really the entire time we've been working with most tooling, unless you knew how to code quickly, was that all of your stuff had to be I'm interacting with a paper prototype or a click through on Figma. now I can just have interactivity with full code fidelity. And to me, that's been a huge benefit because I really also consider myself as, I am I'm not a great visual designer. I definitely just use patterns that I think look okay. But I consider myself a much better interaction designer. And that's the thing that I'm much more able to do now than I ever once was. But I basically do need to push it up to Hi Fi to be able to prototype those interactions.

Dan: Well, you definitely get the sense of things like timing. You get the sense of all the edge cases. You get to feel things when you have the live prototype in your hand that you just are otherwise inaccessible or extremely hard to imagine. Maybe if you're a ultra genius designer, you can think of all the different states of everything. I know I can't. So just even being Well what happens if this is turned on and that is turned on at the same time? What if that and that and that those kinds of things are very hard to keep in your mind as a designer. But easy to pick out when you have the high fidelity prototype that you're playing with and you're Ooh, that feels weird or I guess we forgot how that might work and Yeah, I think that is a that is a huge boon that this provides for sure.

Nik: Yeah, you remind me actually when you said how does something feel, there's a great video of Aaron Draplin, the famous logo designer, Field Notes, and he has a video online where he shows a logo design

Dan: Mm-hmm.

Nik: challenge that he does. And I remember one of the scenes that always sticks out in that to me is he he's sketching and he's moving and he says, okay, yeah, now we're doing this and now we're feeling it. Now we're feeling it. And I was like, I know I know what he means as a designer. It's actually really hard to explain what it means to feel it. I can't actually totally convey that to you. Maybe you can, Dan, but I can't convey in totally what that means. and actually that's one of the things when it comes to interaction that unless I have some way to prototype it and some way to actually experience it, I can't get to that state of I'm now I'm feeling it. And this is something that I am able to do a lot more. with these tools is I'm able to sit there and fiddle around with all kinds of things to get to that really so I don't mind sketching in Hi Fi because I can work on sketching interaction as opposed to sketching visual design.

Dan: this leads us into our second big theme that we've been talking about all year, which is taste. is taste that thing when you know like now I'm feeling it, that's taste. That's what people have been talking about. That's what people think is our moat. I know that I am a little bit more dubious about that, but I did like what MC Dean when we had her on the show talking about how you're gonna be able to get some of your actual taste into things like skills, but you're not gonna get all of it. You're gonna get at most, fifty percent of your taste somehow codified into a skill. And you're not gonna get all the other stuff, your lived experience, your eye, your the strangeness, your life experiences, all that stuff that comes from you, that comes from being a human being. I feel this way both in design and writing too, where it's like, Well that sentence doesn't feel it doesn't feel right and sometimes I can't even express why that feels wrong. but I know how to fix it and I can massage it till it feels right. that's where I'm at with taste right now.

Nik: I still wonder how much more we can capture. I will say there are students in my group who are interested in this question of being able to capture taste and expertise and tacit knowledge.

Dan: Is this a is this a Pokemon thing? for their capturing with the little balls and stuff? Okay.

Nik: No, it's not that. no, I think it's more boring than that. It's how do we represent people and their expertise and how do we elicit information from them and stuff that, which is a lot of interactivity, but So we're looking into this. I don't think we're there yet. I think there's some open research questions in it, which for us is fun. when I run a research lab, so we need those questions and things. I definitely see it among my students who have a developed sense of skill and taste. They are able to produce awesome outputs, things that I think are great. I can recognize the taste that went into it. And when students don't have it, I can see that it's, yeah, that looks kinda everything else or I can probably tell you use the skill. It's weird that I can tell that, but I can, or I feel I can.

Dan: You can. Yeah.

Nik: So I don't know, th

Dan: I de I can definitely tell when someone I can't tell when someone has done something of their own or revised it or tweaked it and a skill. But I can tell when something is raw AI. or a raw even with a raw skill, I can still tell that it is that. That there's been no human polish on it. And the question is, will that last? Or is that something that will always be there? That will always need that level of refinement. per se. That that the AI is gonna be right now it can get everything to I don't know, seventy percent. And with a skill you can get it to seventy five percent. Will it ever get it to eighty to ninety percent? that How high can you go?

Nik: I'm not sure.

Dan: And do and I guess the other question is how high do we want it to go? And that's I think leads us into our third big theme that we've been talking about is this idea of cognitive surrender. How much do we want to give over to the AI? How much do we want to have the AI always do we started by saying, Yeah,

Nik: Yeah.

Dan: You let Fable run ahead for an hour but you felt weird about it. did you feel you were doing too much cognitive surrender there?

Nik: I'm not sure. I'm still thinking about it. Actually, we were talking earlier today about cursor. Cause actually, some news, Jenny Wen, former designer at Anthropic, is moving to Cursor in the near future. And I loved working with Cursor. There was something about the way I would develop and design with Cursor Very fast, back and forth, function by function, and it was hugely empowering. I've since stopped using Cursor, partly because Claude is very good and you have to use Claude code to use all your credits. There's some lock-in stuff they're doing there. But also I am able to move a lot faster. But I am having to sort of say, okay, I'm giving up a little bit of this feeling of kind of being in really steering. Before I felt like I was steering, now I really feel like I'm directing. and I'm sort of giving orders and letting it happen. And I don't know how I feel about that. I do feel sometimes I'm surrendering more than I'm just sort of offloading. but I do always feel that I can come back and go, wait a minute. So actually, to do my weekend project, I did sit for about two hours working back and forth on a product requirements document. And I did not just say, cool, you write the whole PRD and I'll just accept what you want. I actually read it line by line and was what are you doing here? This doesn't make any sense. Why have you made that? And then I would come in and be no, change that. Don't do that. Then there's other stuff. You're no, this is good. Hey, you made a good architectural decision. that makes sense to me. I would also do it that way. but there so I'm I was fine being in this really fast back and forth in regards to the requirements and the spec. But then yeah, when it came to the code, I was nope. You take it, but I don't know how different was that when I work with one of my students and we come up with a plan and then I say, Great, go and build it and they come back and go, Here you go, Nick, I built it and let's try it and then I give feedback. It's actually not that different than that experience.

Dan: in my case, I'm like, well I have never written a line of Swift. I couldn't do that at all. I'm not de-skilling myself because I could never do that. I don't care that Claude handled all of that for my project because I was never gonna be good at it. my only B in grad school was in a coding class and that was extremely generous of the professor. Shout out to Golan Levin, CMU professor, who took pity on me and gave me a B

Nik: Yeah.

Dan: there's certain things I don't care that I am reliant on it. I am perfectly fine offloading that stuff. It's the stuff that I know about That I then do care about the polish because I can see it and I can feel that it's not as good. Now maybe someone who was a Swift coder could would look at this code and be like, this is garbage. I would never do it this way. but I I don't know that. So I don't know what I don't know. And I'm fine with that. maybe in bigger things or in critical things it would be a completely different story. But for smaller stuff, I don't know. I'm having trouble caring that much about that cognitive surrender. But with things that I'm good at I'm not gonna use Claude to bang these things out because why? it will take me just as long to describe to Claude or Gemini or Chatty G to do it but I still think that as a general concern, I am still concerned about this because for the exact reason we talked about. it is too easy to just accept what the AI gives you and be like, this is fine and not do that review. Especially if you don't know about it. And I think that that's that is something that we need to still look out for. It's and it's just getting worse. I don't know how to fix it.

Nik: Well, as the things can do more and they're more capable, it's just so much easier to allow yourself the release of surrendering. So I was doing another project this weekend and I needed to do some back end stuff. I required a database and I needed to do some scraping and some processing of that data. And I started by saying, Okay, I'm gonna sit here and really steer it, make some decisions. And at a certain point I kinda got tired and I was I don't know, it's doing a good enough job. And I just said, you what, I'm just switching to auto and I just let it go. And it that was one of the things that made me feel weird. But again, it did fine. maybe it does fine until it doesn't. but also this was a personal project. This is something that I'm not even sure I might share with anyone else. It's just code for me. It's me making my own, little apps, true end user programming and For that stuff, yeah, maybe it doesn't matter. You get the thing you want at the end and you're like, great, I have the tool that I wanted to make and I'll move on with my life. And if it breaks, well, it breaks for me.

Dan: One of the things that I tried to do with my little project this weekend was to Not touch Figma was to do it all in Claude, so I did it Claude code, Claude and Claude Design So playing around with that overlap of the three playing around basically with what our fourth big theme has been about canvas and direct manipulation versus prompting and code and trying to fit all those together and how good could I make it just from using one suite of tools and I have to say I did at various points really miss not going to Figma to do things because I could do things so much faster and quicker in Figma and the canvas on clawed design is not quite there yet but being able to do those kinds of tweaks discuss and then have it back and put it back on my phone. it's pretty cool. And to me that's been one of the just the big major shifts going right now. I feel we're in that world at this moment where the canvas and code are converging and I think it's I think it's pretty cool.

Nik: I think that it's gonna be interesting to see where both paradigms end up in the middle somewhere and who sort of figures out what works well and is really effective for designers. And I still think that it's open, I know that We have talked on the show a lot about wow, and Cloud Design, that was pretty impressive, what they put out and what they're doing at the same time, Figma's coming up with a lot of stuff and who knows, there may be other tools that come out of nowhere because again, there might be a team somewhere literally who's just been just burning fabled fable tokens last two weeks and all of a sudden we're gonna have some crazy tool that we haven't even seen yet. I have no idea.

Dan: Ha I guess this begs the question will it even matter? Are we just gonna be designing for agents from now on? that's our next big theme that we actually started talking about I think on our very first episode of this was that Your next user may not be human. This idea of MX, the machine experience. So agent friendliness is the new UI all the stuff around MCPs and all those things that are just happening in the background where I'm just hey Gemini, get me this thing, go do this thing or hey Siri, go do this. All those things are coming out and the question is, what's the design when nobody visits your site,

Nik: It's fun for you to bring up the topic of designing for agents and that machine experience because while we've talked about it a lot, I actually feel like that is not something that has moved a whole ton in the last couple of weeks. MCPs exist. I will say that anytime I use something and it doesn't have an API and or MCP, I'm a little like, if it's a new tool that I wanna use, I'm wait, no, no, no. It has to have this because I want to connect it to my agent because I using it. And I I'm I wanna automate things. there's little things that I wanna automate. And having the MCP lets me automate certain things and then I do other things. And so I I like that. But that that's me. I might be out on the edge on this. But there hasn't been a whole lot else that I've really found moving us forward in that thinking more. I don't think I've seen more stories for it. I feel like we've spending a lot of our time, more on yeah, maybe the canvas versus code discussion, the developments there. yeah. So I don't know. it'll be interesting to see maybe in the next year or maybe we're reaching a place of stability there and it's just a matter of groups having to implement MCPs and other forms of this into their applications. actually when was the last time you heard about OpenClaw? I have not heard about OpenClaw in what feels like months.

Dan: Open Claw was the tip of the spear and now we're at the floodgates. We've got, clawed cowork, right?

Nik: cowork is try is trying to do some of those things for you, maybe.

Dan: Right, what's the one that OpenAI has now too? right, other people are now going after that exact same space. I don't think that the state of the art has moved.

Nik: Work.

Dan: Although the models have certainly got better and the tasks have certainly got longer and you certainly can deploy many, many agents now for long periods of time. It'll cost ya, but you can do it.

Nik: I was gonna say that it's really costing you a lot more too. That's actually the thing that I think could be really dangerous with, an open claw instance now if you put it on one of the bigger models, is that it just burns tokens, which is pretty much that's, money out of your pocket every time it does something. Whereas the companies where you can have subscriptions OpenAI, Anthropic. and you're using their tools, you're all just within your subscription. You've already paid them for it. So now you feel okay, I can use a bunch of stuff there. I think that's maybe that may be one of the reasons why we don't hear as much about it. especially with some of the bigger models.

Dan: I think this is something to watch in the second half of the year for sure. our sixth big theme was all around skills. We talked about Impeccable in episode one slash delight. And then we had the Queen of Skills, MC Dean on back a couple months ago talking about the hundred plus open source skills that and she's added I don't know, fifty more since the time that we talked to her. And yeah, just the rise of the plain text markdown skill is Something that I think is really cool. I'm curious to see what Figma's gonna do with this. they announced that it was coming at config but didn't show us anything. So I'm super curious about where this is headed in the next six months.

Nik: Yeah, and I think that there's so many skills that are out there now. There are now starting to be repositories that you can download skills. Actually, you had shared one with me, which basically had created visual design skills of like hundreds of real websites which felt kinda weird to me. at first when I saw it, I was like, cool. This is the new WordPress template downloader sites that I used to go to when I was younger. And people had made custom templates and stuff. You could even sell them. and then I looked closer, I was no, They've just they've literally just ripped off all of these, real products with teams of professional designers who put a lot of work most likely into their stuff. And that didn't feel great to me. And I was but yeah, it you can just download the skill now. And I have no idea how well they work. We'll link to it in the show notes so people can go take a look at this. it's weird. I don't know, as a designer, I don't even really want to touch it. if you work for this other company, I might be inspired by your work. I'm always inspired by the work of others. So I'm always gonna be taking in saying, there's something about this that I like. But I don't wanna effectively try to reverse engineer your design system and then scrape it and put it into my thing. I don't wanna do that.

Dan: It definitely has the ick sure. That even though it's technically probably fine, I legally fine, it's feels it feels morally wrong and just strange. it's interesting from an academic perspective in that A it can be done and B it could help you if you're looking at these like what's the Apple style in markdown format? That's interesting. But to actually then deploy the Apple style directly, ugh, that feels bad.

Nik: Yeah, one thing that it might be useful for, to your point, that I have to see if they work well is the actual structure of those files and understanding everything they put into it. Because that's where I could see if you haven't written a skill before and you can just point an AI agent and say write a skill for this. But actually if these have been somehow curated or they've been tested or something that, you can see what goes into it and then build your own skills by learning the format. That's cool. I don't think there's anything wrong with that. And then you just design your own stuff or you maybe give it some examples and then say, hey, here's a Figma mock-up that I have. Please extract the skill in this format. That could be a really useful tool. But if we have all of these skills now, and I can just save these skills in the repo. What does this mean for what our jobs are? And actually specifically this goes to one of our next themes has been the sort of changing role of our work, especially for junior designers. if these skills can be codified and they can be saved, they can be shared, it means that they can be saved within our companies and our organizations, that's awesome. They can be documented and they can be in the repo. But what does this also mean though for say the work that we do and especially the work that we would do as a junior designer? Because many of these skills might be doing some of the things that a junior designer might do. And this brings us to our next theme, which is what is the changing role of that junior designer? And are we even gonna have junior designers, are we hiring them anymore?

Dan: One thing that we didn't talk about on this show but got released in the last couple of weeks is there's a state of the technology industry research report that comes out every year. it's affiliated with Lenny's podcast and One of the questions they asked is would you encourage people to join your profession? And a shockingly large number of people in the tech industry said, no, don't don't do it. particularly designers and researchers were the two that were like, I wouldn't if I was you. And I think that just has to go around the uncertainty of, yeah, are these skills taking too much of my job? I Don't believe that. I think that jobs are not tasks. I think that What the designer does is far more than just put the pieces together, put the right prompt together. any more than I thought that the designer's job was to push pixels around in Figma and make a UI. I just I don't think that that is true. I think there is a junior crisis 'cause we see it. We see fewer junior people getting hired or it takes them longer to get hired than they would have in the past. we have a front seat to that. it feels like the job is changing, but it's not going away and I wish that more companies and teams would know that. going back to Jenny Wen, that crack new grad, people that are hungry and fast, they don't have that legacy process, that are AI trained. those junior folks who are, really AI natives who can work really well with your senior designers who know your context really well or and your user researchers who know your context and your customers really well. Man, that seems like that is such a great combo that I wish more people would pick up on.

Nik: Yeah, I agree with that. And I think that one of the things this conversation reminds me of, I was reading Corey Doctoro's blog post on reverse centaurs. And so the idea behind the centaur and reverse centaur is that, as a centaur, so you have the body of a human and the legs of a horse, I am able to augment it and do more. And I feel like actually when you and I talk about our personal projects, we're talking about being centaurs. that this is stuff we've never been able to do. you never programmed in Swift, you weren't gonna learn it. Now all of a sudden you can make a Swift app and actually realize an idea that you had, that's pretty amazing. And you're choosing to do this, you're also choosing to do it willingly. The concept that Dr. O talks about as this reverse centaur is no, no, no, I am making you do this. But also that it is more that like you must put in the work, you must take the fall, you must be responsible for all the things the AI does, but you have to work as hard as you can to make sure that the AI does all these things. And that's actually maybe a little bit of like when I'm feeling. this I have to keep shoveling is I'm getting a little of that. And I think within companies, right, there's a big risk here of forcing this on people to meet expectations that you're setting higher and higher and higher because you believe that this can be done, even if it's still somewhat un proven that we really are Creating lots of business and user value. We're creating lots of stuff, but again, there's still the question of are you creating the value that it demands? And this is something that I think I do worry a little bit about, that you might get a little bit complacent and be like, well, I can use the AI, I'll just use the AI to do my own thing. I'll just centaur myself almost into becoming eventually a reverse centaur because you don't have anyone else helping you out. You don't have that team to work with you and now you are stuck. I don't know, maybe that's a little bit of what I'm feeling. I was I didn't have no one to talk to, I had no one to hand things off to. I don't know. It's just a thought here, that I could imagine, comp companies and teams starting to wrestle with.

Dan: The other thing from that state of technology report was that people were less afraid of AI taking their jobs and the thing that they really hated the most was doing more for the same amount of pay and this yeah, this reverse centaurism where it's okay, well we have to keep feeding the AI. rather than three PRs a day, we have to close thirty a day. you can close thirty a day? Can't you just build a tool that would close sixty a day? it's that running in place, so someone compared it to that your car is a neutral but your foot is on the gas and it's that feeling of we're running, just to keep the machine going. And yeah, it I I hope that we start to see a lot more value metrics coming out of all this. yeah we're producing a lot more but where's the where's the value? Where's the beef here? we should be seeing all these amazing gains in this last six months. All these super cool new products and super personalized and hyper efficiencies and I'm not seeing that. Are you seeing that?

Nik: No, I'm not sure I'm really seeing it. And from a work perspective, I'm not really seeing this beautiful future that I had thought I had imagined of I'll be a relaxing. I'll be working. I am, I guess, working on my own things, and maybe this is a problem I have of I to create stuff. But I'm sitting in front of my computer a lot more on the weekends and not going outside. And again, that might be a me problem.

Dan: Yeah.

Nik: But it is this weird thing. I maybe I've just pushed myself into this thing of well, I have to do more. I have to do I create more. And I don't feel we've gotten to this point of it's not a calm interface. it's not a interface, it's not an interaction to me that feels this calm, peaceful way of computing and a peaceful way of creating something. Unlike maybe when I'm in say the actual studio and I'm drawing or I'm woodworking or I'm molding with cla like those experiences to me can be almost meditative. There's nothing meditative right now that I feel and actually in some ways I feel I've lost some of the meditative or flow based feeling that I got when I wrote code by hand. So we're not there yet.

Dan: I get that feeling when I am revising. when I have taken output and I am polishing, but that is also the piece that is the most like my previous way of working. So I maybe maybe it's that. When I am generating or doing a lot of prompting and stuff that, that's when it feels more frantic than it does when I am immersed in okay, I've got this raw thing. How am I gonna make this better?

Nik: So our last topic, and actually I think this is a great way to end because you and I we're working on our own stuff, but we're coming up with ideas because we see something in the world that we'd like to see different, that we'd to see better. I love your app idea. I love the idea of being able to recognize bird song without pulling my phone out. Actually, from an experienced design perspective, I think it's lovely.

Dan: Yes.

Nik: But you had you still had to recognize that you didn't sit down, I assume, and chat with an AI to say, help me come up with product ideas. No.

Dan: Nope. Yeah, no. my girlfriend was pulling out her phone and holding it up in the air, trying to listen for a bird, and I'm like, This seems crazy. Can't you use hair pods to do that? I saw a small piece of friction that could easily be removed and make the experience very pleasant that as you were just sitting around or walking, it would just tell you, yeah, that's a cardinal that you're hearing.

Nik: And so this is the theme really, user research in the AI era, I think that is undervalued. I think actually we need to be doing more user research. if you are a company that's out there, that you're building skills, you're getting a new design production system in place, and your design team is able to produce more prototypes, more code. to a high fidelity, to a fidelity that you feel is good, then great, come up with great product ideas. That's still something that you at least right now I'm not seeing a lot of automation in that space. And so I think you need more user research.

Dan: we've had this idea and we've talked about it here with multiple guests. we talked about it with Tom Rimmer, Tom Groendahl, Shane Johnson. Yeah, this context manager is a great thing that would be awesome for user research to start. feeding into, if not becoming, that role where they understand and can codify some of that understanding into these systems. I think it's a major opportunity for companies that is being overlooked probably because a lot of those roles got cut and were never replaced. I think it's actually part of a problem now where the thing that people are looking to do feels is focus on the people that are the that are that top the people listening to this podcast, that top ten percent that are extremely AI focused and in the know about what's happening with the latest and greatest and There's a huge ninety percent of people that are not using these tools in the way that a lot of us are, because the tools are still too hard to use and they're still too opaque, and there's people still under get the value from them. And these are all things that research and design should be focused on. How do we democratize this technology to a broader group of people rather than chasing the latest and greatest models and what those models can do. If the models just froze today, we would still have so much work to do to bring everyone in and show what these tools can do. And rant.

Nik: I think that's a great place to end it actually. it's been really awesome. This is episode twenty six, right? We've been going now for over six months, half a year. we're excited about the next half of the year. We'll see what happens. I'm sure there's gonna be some big announcements and all kinds of new things that occur and we'll be here to discuss it each week.

Dan: Half a year. Yeah. Yeah, and watch that AI brain fry. We'll see you next week.