The Psychology of AI-Driven UX

[Music] Sophie, welcome to the Stutterwork Podcast. So glad you're here. Thank you so much for the invite and very happy to be here. I wanted to make this episode because in one of our last episodes we had someone from Uber and one interesting fact was that he said that the limiting factor when it comes to deploying you software, deploying new features is actually not so much the capabilities of the engineers, but the customer's capability to adjust and adapt to an ever-changing piece of software, piece of product or service. So I reached out and I wanted to make this episode with you on the psychology behind that and trying to talk with you and maybe look into the future of how product management and product lifecycle might look like in an age where you access maybe overwhelming people using our apps and programs. Yeah, I mean I think it's a fascinating topic and I mean I'm not an expert in psychology that's just a disclaimer. I do find really interesting, especially as a researcher that you have to understand those basic principles and then also when you build four or with AI, you know, I think there's a distinction there. You have to really understand what's happening in humans brain and we all go at different speeds. So yeah, I think it's going to be a great episode. So next time. Okay, then let's jump right in. Can you give us a brief introduction to who you are for everyone who does not know you yet? So my name is Sophie. I am French. If you can hear from the accent, I'm currently based in London and I've been working as your ex-researcher for the last four or five years. So I have a background in chemical engineering and kind of came into tech a bit by accident if I can say. And yeah, I've worked across startup enterprise done B2B, B2C and right now I work on internal systems which is also very interesting, you know, compared to more like traditional UX. So yeah, that's it about me. What is internal systems for everyone who does not know what that is? So if you think about, obviously like if you think about normal company, you would try to sell good or services to like, you know, other companies or consumers, internal systems or I would say mostly for large, I mean, really large company. So you're client as within your company? Yeah, exactly. So my, my user is at the same time my colleagues or other people. So yeah, it's all very interesting and different constraints as well. That space. What do you think is happening out there in the UX world? We saw a cloud launching a couple of things where the AI is already anticipating of what the user experience might look like and designing user interfaces. And what does your take on the speed that we gained in that field regarding to the capabilities of the human brain? I'm going to be very honest in this episode, you know, in terms of, so I think there is really like productivity paradox because at the same time, you have all those, you know, obviously like big AI companies that release tools at amazing speed, but at the same time, users, you know, tech users, we should be at the forefront of that. I feel like we all already feel behind because there is so much to catch up on. And that's literally our job, you know, that's our space. So when it comes after to implementing those experiences with the users, you know, there is a distinction, I guess, where if the user is, you know, actively using AI and needs that or is AI kind of the background that enables users to have a better experience. So I think there is two parts there, but I would say it's going very fast, but is it making us already productive enough? I'm not sure, you know, I think there's a lot of things that people like say, you should be doing these, you should be doing that, but at the same time, the technology, the infrastructure doesn't follow, like we're talking about dinosaur tech, you know, where you sometimes even putting an API is already hard. So how after that, the company saying, oh, we need to put AI on top of that for what reason, you know, sometimes it's debatable. But yeah, I think there is really like a kind of paradox there, where we feel like it's going super fast and we should be on top of everything, but at the same time, the reality is different. So that would be my kind of high level view on that. I think that's interesting. We should dig a little bit deeper into the two paths that you did you just came up with. So one would be the UX, where you just have basically, you're using AI as the end user. So you might just have like a prompt bar where you type in what you actually want. And the program is providing the AI with enough context from your account, probably, to give you the response. So there's no, it's not like you're clicking through an app or you're clicking through a program, but you're just interacting with an artificial intelligence and that would make all the things for you in the background. That is one thing. And the other thing is maybe interfaces stay like that. And user actually know what's happening in the background. I want to be part of this. But on the development side of things, people use AI and ship much faster and bring on change to pieces of software more frequently through it. Do you see a trend in those directions? And which trend do you see? Yeah, I think obviously if you think about front-end development, they can ship faster. Backend, I guess it's still a bit more complex there. I see designers being more hands-on with coding. And you know, it facilitates the hand-off with developers. Obviously, as well there, you have kind of things increasing. But at the same time, you know, maybe that speed makes everyone speed. So when it comes to actually sitting down, debating, ideating, etc. I think stakeholders can think a bit like a shortcut. Oh, there is AI. Let's just try to mock up something right now, which in some cases might be useful if you need a visual, if you need kind of like a starting point. But I think we shouldn't miss that first part, which is very important in design thinking, which is actually to understand space and empathize with users and really try to come up with ideas and test them as we go rather than, you know, solutionizing to early, which was already there. But I think with AI's, just made things worse because everyone can design at the end. And or at least it seems like everyone can. Is there a certain limit to the amount of change that you can actually ship to a user? I think if that's the moment to talk about psychology and, you know, kind of those principles that, you know, whether you design with AI or not, you should follow. But the first one is of the cognitive load, which, you know, we know how overwhelming it is to sit in meetings for five, six hours. And there, you know, it's just already cognitive load. But when it comes to actually, if you think about a website, there are different studies, you know, it varies, but on average is between, you know, three to five, three to seven items that the kind of working memory can. So if you have like five elements, it's already a lot for the user. So obviously, like cognitive load, you should kind of be careful about that. Change fatigue is also something, because I think we as human, we rely a lot on kind of our habits, our reflex. Even in our everyday life, we all have our team, we all have our habits. So there as well, if the user goes on the website and, you know, kind of new by how to go. And then it's completely gone. And they have to adapt constantly. That is not great. So I know there's a few like kind of UX principle that we should all try to follow to mitigate that change fatigue. And, you know, at the end is disengagement. If users have to constantly relearn and, you know, they don't have the capacity or the, the effort to do that, then they're going to go somewhere else. So ultimately, yeah, I think it's, you should be careful about releasing too fast or at least bringing that in a very user friendly way. Yeah, I very much remember when, I don't know which windows updated was, but they changed the whole design. I think it was, it came with a 2000s windows 2000. And it was just so different that by the time that I really opened it up, it was just feeling very, it didn't feel familiar to me. So I really very much had to explore it all over again. And then moving from 2000 onwards, I didn't really get in touch with it the right way. So I kind of switched to Mac and stayed there for the longest period, right? You're still there. And yeah, and I'm still there because it still feels like the first time I opened up a MacBook, like from the, from the UX perspective, very much similar to to what I saw many years ago, right? And things that are familiar to me. Yeah, I think Apple have been really good in terms of making things intuitive. And I think that's, I mean, that's why, you know, they have a really big fall in market. That's I think they have managed to, even if they add new features to keep it fairly simple, it's guided that anybody, whether you're like a tech savvy or not, that you should be able to use those. So yeah, definitely that's one of the best examples. - And one part is actually working with the product, like clicking through it and doing something. And now when I'm thinking of my workflow, I very much like go away from building something directly on my MacBook instead I moved to just checking the result of the AI that built something for me. It's a very different thing, right? - And I think there is a shift there where you, before you know, you kind of finish the flow and then you're done. Now it's more about degenerate, but you have to validate. So the user is not doing, just doing any more built of verifying. So sometimes you could also think, are you accepting the output a bit too quickly because you know you can't be asked to like, you know, like it rate your prompt for a long time. And at the same time you also have the trust, because if you have to constantly check, because you know, AI especially with things like, you know, basic charge, GPT, you have to experiment with it to test how it works for you, what output looks good or not. So you're so kind of use that trust in the system because you constantly orbit on it where like is that satisfying enough for me? So yeah, it's a bit of a shift there as well. - Yeah, what happens when the AI completely takes over my workflow, let's say, and I can't really tell what is happening in the background. While before AI, I was able to sort of, I click here and it prints a letter A onto the word document, right? And now I'm moving away from that. And somehow I give it a prompt, and depending on the model, depending on the context, depending on my settings or my prep prompts, on all of that, it creates, it guesses something, right? And creates that. So I very much don't know the result of my actions. I'm always just like throwing a dark blind folder, right? How do we, is there a way where we can design a user experience? So a user has trust into what they do with the piece of software and into their own workflow. - I think one of the basic is transparency. And you know, you have to define what level of transparency of what is happening in the background is acceptable for. The user, someone, well technical, might really want to have access to the code. Whereas for me, I just like to know, you know, if I analyze an Excel table, what kind of are the big steps that, for example, they've done to arrive at those results. So you have to define that level transparency, but I think always give, I think one of the key things is to design for recovery. So being able to edit, being able to update, without it being too complicated. I remember at the very start, when CharGPT was there, you know, you didn't, it was very much, you write, I give you an answer and you have that chat and that seat. I remember talking about what I do, AI would look for UX or especially as a UX researcher. And I think I was already talking about those ability to being able to analyze different type of media, being able to retry more easily, and with the AI taking into account all my changes. And not just the last message. So I think designing for recovery in general and for flexibility, I think that's where we are heading at, where the system is very adaptable and can anticipate and still allow the users to be in control, rather than the user having to just open another, you know, chat because I'm just fed up. And I just want to start from scratch. Yeah, I've been there, done that, I can tell you that. So just today, I was like trying to get a document ready and Cloud didn't really upload my company's logo onto it. So I really find, I took like six iterations onto the skill in Cloud to really make that shitty piece of software put the logo where it's like the right logo. Because it guesses the logo, it looks at it and then it anticipates why it's somewhat a parallelogram and just put some type of parallelogram onto the document, right? So I'm wondering what the trajectory is that you're seeing at companies. Are they already like creating user experiences with AI and do you think it will ultimately be the trajectory to have an AI design? The user experience from scratch or would it still be like a human endeavor to create such a thing? I think in all the AI workflows that we are building, you always need to have the ability to have a human, manually intervining. And I would say at any stage that they want, because also for some people, AI will work well on step one and three and for others on step four. So you can't already generalize because you all have our own way to function. And especially one output that they give me, my satisfy mirror as my colleague, my might not be happy with that. So I don't think there is a uniformity yet that is coming. I think there will be principles that will emerge from that. And I think we are already starting to see online, AI principle, our designing with AI. And also a lot of the things outside of the technology will be a lot about what is ethical, data privacy, et cetera. So I think it's just really there's a lot of discussion going on. I don't think there is a clear direction. Whereas is AI going to design everything? Well, on paper, it could. Is it going to be a good thing? Probably not. So-- Well, it depends, right? Is it a good thing for the one creating the UX? Is it a good thing for the company who hired the UX guy? Is it a good thing for the AI company that is just selling your refined quality tokens? I do think-- A lot of things going on. Yeah, I do think-- I like to think about AI as a new tool. When there was no internet, I think people before they were working, like looking in books and maybe traveling more to meet people in person. And then there was internet. So-- And now a lot of things are AI-driven. And I think it's a bit-- obviously, as from the marketing or branding perspective, you have to put that. But I think now it became kind of common knowledge that most of the product might have some kind of AI. So I think after it's how-- it comes more to the decision-making of the company. If they are-- because you still need to involve your users, even if you design with AI, even if you code with AI. That, for example, I'm not worried about my job as a UX researcher, because you always need that expertise. And I mean, after I try the AI-moderated interviews or surveys or things, and even if it can speed up, I personally find it a little bit weird. It's just doesn't feel right. And I think the same for designers, they started by drawing and being on paper and just having sticky notes. I mean, that's the image of design thinking. I think it's still going to be there. It's just going to look different, because you will have AI in some of the ideation, for example, or the analysis. It will help you put things in place more quickly. But yeah, I don't feel worried about that. And if the company wants to get rid of designers and UX people, let's see what's happening, because I don't feel very confident for them. Yeah, have you seen a trend in a lay of swan that becomes to designers in UX research? It's like, let's say, with software developers. I think everyone is more or less at risk. We're not just about AI, just because of the post-COVID and restrictions on budget. And everything that's going on in the world. And I think I see a lot of designers or UX people that are all really worried about them having to be absolute experts on AI. I have to have tested all the design tools. I have to show that there is AI everywhere on my process. Personally, I seem very old school. I do rewatch my videos. I do take my notes. I do spend a lot of time reading. And I wouldn't trade that off just to have a chart just giving me everything, because there is a lot of pleasure. And that's where you build your empathy and your judgment. In those phases where you sit down and you think. And not necessarily just replacing everyone by AI would be good. So yeah, I would say it's been a bit more competitive, more layoff. But at the same time, it could maybe next year could just increase a lot. And then two years after go down, I just think it's just more unstable, but it's not impossible. To-- Let's go a bit on the good side of things. You already mentioned a couple of things where AI is really helpful when it comes to designing this experience. in your current workflow that you're having. Can you give us a bit of a perspective on what a good AI workflow would look like when it comes to designing user experience, any good practices that emerged already? So I'm going to talk more from a research perspective. From a research perspective, I've tried AI at every step, so from planning to designing the study and analyzing and presenting. I would say it helped me the most at the planning stage and the presentation stage, mostly because sometimes you just have so many information. As a researcher, I love having a lot of documents. I love having the notes from a lot of meetings or some people have said. So then it's really helpful to compile that and make sense of the data. Especially when you're on your own a project, then you don't really know the space or it's quite complex. So that definitely helps me a lot and probably saves me a lot of hours. And do you do you use AI to organize the data or do you just throw it in I don't know, into notion and have the AI fetch the piece of data that you're looking for? So I can't use it at my current role, but I used to enjoy a lot, not Google, not Google, but KELOLM. So obviously you upload and after you can ask questions. Yeah, it just depends on what level, you know, sometimes I just want to have a high level understanding of what's happening just to catch up. And then, you know, when we go more into defining the real objectives and the question, that's where we get more targeted. But I still think about it a lot myself. And then I like to compare what AI come up with. And sometimes gives me like an angle that I haven't thought of for. Just, you know, an idea that I like to bring up to the team. So yeah, just varies, you know, but always stay critical when you use it. I would say that's probably the main key part. Yeah. And let's say you created a product or a service using AI and then you ship it to your customer. Is there any change in the metrics that we should look at when it comes to how people interact with our product? If we designed it with an AI or if we produce developed it with an AI. I think that how do we really get the data behind our vibe code at Slop that we put out there each day. Yeah, so I have, I took a few notes, but I would say they are, well, I would say the traditional metrics. So, you know, time on task. The support volume, you know, after you really something retention. So I would say all those kind of traditional metrics are still good to keep in mind. After I think there is, they are a metric that a bit more AI specific. But one that is very tricky to measure, I would say is trust because as we talked at the start trusting the system because you have to verify the output. You know, there are methods to measure trust, but I think it's very subjective. So I think that's where, you know, we might still a bit more work on defining those metrics and establish them. But then even just the rate of, you know, undo edit on the AI chat. Yeah, like all those kind of micro interaction that the users might have just to fine tune the prompt or. You know, either is because they're really happy and they want to push the reflection or maybe it's because the system is not working well. And that's where you ask research user interviews and after you understand why. So there is still this complementarity of quantitative metrics. I think there are some that are more AI specific, but you still need that qualitative lens and, you know, involving the user as you go through. And then you can't just comment build you know a whole product and just hope that users are going to learn because he didn't work before I won't work now. You already mentioned the trust and let's get back to the for a bit. Is there other like patterns that you could use to actually increase trust anything that you've seen. What are you doing research? Like you expect that people could follow to increase trust when people attack with your product. Yeah, I think one is for example, just to keep the core flows stable. So what users are familiar with keep those elements there. Don't change it overnight. Because it will, you know, they will recognize your system. So even if you changed. And then it's a bit easier and it's more familiar. Then I would say no massive big change from one day to another and the product looks very different because you might be like, oh, I, you know, what's happening overnight. And whether you what do you put the line? So I get it that you won't change shouldn't change your whole product from one day to another. So if I change let's say a certain feature or like is it how can I identify what a good amount of change would be an acceptable amount of change. Now you have to test with the users, you know, it's very dependent on on how much. But as a UX practice, you should explain what changed. So you know, having whether it's help or maybe like an onboarding if really the feature works very differently. So something like we moved your bar from here to here and you can now type it in and you get everything you want. And then have like a button where people like don't like the new design click here. You get the old design. Yeah, some some have done that, you know, recently I saw in Vanguard they have this old and new design and they're like, what do you want to change. But I think explaining what changed and how it's going to affect users or not. You know, and having for them the ability to serve serve because you don't want them to email you or call you, you know, it's not it's not a good sign if you start to have where it's ex gone. So yeah, being able to also communicate those changes, whether you know it's automated emails or having like onboarding or help help pages. I think all of those things will help users trust the company and be like, okay, they are not just putting that in our face. They're actually trying to get us up to speed. And they're not tricky. I mean, there are companies out there not naming anyone except Microsoft, who is who's really dog feeding everyone their copilot at the moment or better say they're like, I guess, 48 versions of copilot. So it's not just just recently, but it showed that it had like, I don't know, I mid size two digit number of different products called copilot something right. And I was wondering whether there is like a good UX practice emerging to really know when to use AI in a certain functionality or feature and when to just leave it out and leave it as it is maybe or just change the interaction but you don't actually needs implement an AI behind that. Is there a rough guideline on that? I think we've all seen products that have just shifted and rebranded themselves AI, you know, during that the last two, three years. And I think it's for a lot of them has been very quick. So again, I come back to what is the real problem that users are facing? How can we help them? And then is AI making sales from a feasibility perspective? Is it going to save us time? Is it also going to benefit the user? And also what harm is there towards the user if we use AI? Like, you know, what should be one them about whether it's, you know, how their data use or ethics, etc. So, yeah, I would say it's not everything needs AI and that's going to be the same. I think even if you get the most sophisticated AI software, I think some stuff work well as they are, you don't need to over complicate things. And sometimes you're like a company decision, but it's good as you know UX practitioners or tech practitioners to push back and to really push them to be like, do you really think we need AI? Is there not an alternative solution? And sometimes it will be simpler and it will cost you less money. Yeah, think about it. You mentioned ethics a bunch of times now on the show and when you say users not aware of the ethics behind applied. Can you elaborate on that one? I think I'm not, you know, I think there's a lot of people that talk about ethics with AI and within how, you know, the data are being used, for example, or how it can arms, harm people because I'm not aware that their pictures are being used or just you know, even on social media, what people ask AI to do, I think it's just insane sometimes. So I think yeah, we have to be very careful about because now fake news and misrepresentation, all those things are quite frightening. I think sometimes when you see what's happening, so it's also being mindful about how we use AI and how it's going to benefit us, but also the users in the first place because we always about the company company, the benefit for us, we're going to cost time, we're going to cheap faster, actually, what is also the benefit to the user, plus or so, all the environmental aspect of AI, which is another debate. So you think that companies should be very transparent about that and put it somewhere, because it might also be. And in inflicting uncertainty, when it comes to the purchasing decision, when people are on a trial using your product, and you then start to talk about the ethics, about the environmental impact or the data labeling and all of that. I don't think it's about the company having it into your face when you're on board, but it's just being mindful about it as we are so built with AI, just to make sure that the AI model for example, already very biased, and we've seen out of examples how they are biased. So it's more about kind of free shifting and making sure we don't cause more harm to what is already there, but obviously every especially big companies will have their own department who looks after all those kind of data privacy, compliance and everything else. So you don't need to think about it all the time, but being mindful about it and trying to practice the AI implementation as best as we can, I think is already a good step. If every tech practitioner acts like that, then I think we might head in the right direction. Maybe answer a different topic. When it comes to numbers, it's always super interesting to see statistics and such, and do you see a change in trends and numbers, like a measurable impact when AI entered the UX space? When it comes to let's say the cognitive load of users or on the development and designing of UX, what are the things that change metably when AI entered that? So, Oshia, not that expert, I have my notes. I think there is like, I see the working memory, which you know, you can, those kind of, it's a bit hard with AI to say how many items, but I think like, I've seen a study where it's like they look at the, I don't know how you got, like, the brain activity and you can see like, sometime with AI, it kind of is more active. So you can see that it kind of requires a bit more effort from us, but there's been like studies in the last two, three years across human computer interaction or psychology and two numbers. One is the task, the time on task, which you know, with AI and if it's not done well can increase by 20 to 40%. Obviously that's what is time, what is time on task? So you know, how much time is taking you to complete one, you know, if I want to buy something, how much time it requires and with a higher collective cognitive load, it increased 20 to 40%. So if AI by nature requires more cognitive load because it's just, you know, being new, you have to adapt, you have to relearn. So even naturally, even if you bring it in the best possible way, you will have, you will always have this element of novelty. So yeah, people taking more time to complete basically what they want. So actually, when I'm running like my e-commerce business and I want to have like a gentry, purchase processes, right? Even though on paper, the performance of the shop should increase, I should anticipate a decrease or at least like a increase in the time it takes for someone to actually close the deal. Yeah, and after it depends on, you know, we talked about those environments, whether it's like, you know, proper AI tool, whether it's embedded in the backend. So that, you know, I don't recall all the details of this study, but it just shows that in Instant Ball, maybe you take, as we said, you take your time because you want to look at things, you want to check what was being generated. So in a way, it shows that you have critical thinking, for example, if you think about AI chat. So it's not bad, but it shows that it's just an indicator that people have need more time. And then there is the kind of error rate, which, you know, we all know how AI has been frustrating and it just starts from scratch. So all those kind of retry error rates also increase in the last few years with all those releases. So I think it's like just some ballpark, not saying AI is totally wrong. So it's just to clarify that error rate is basically when I put in a prompt, I wanted to do something and I have to redo it all over again. That would be, that would count as an error. Yeah, and the same error is subjective. Sometimes the output I'm satisfied, but I would consider it an error because it's just not giving me the level of detail or refinement that I want. Or it could just be like me not being able to achieve my task. So again, there is all this notion of subjectivity, which I think is coming with the interaction with AI and it's quite complicated because we're all very different and we all have a different experience with AI and, you know, personally, I don't really consider myself super knowledgeable in terms of all the latest AI. I try to keep up with the train and experiment that I can. Yeah, it's hard. It's just it. Yeah, there's shit like let's take entropic for one because that's one I really, really use a lot. It almost seems to be that I have to update it two times a day. It always wants to be updated. There's always something you coming up. It's insane. Like, and just learning about the update seems to be a full-time job. Yeah, because so much things are being shipped and improvements. I assume there must be someone at entropic thinking how can we actually make like make the user actually use our new features? Because like, if I if I'm off-cloud for let's say a week, I'll probably miss out on five new features. And when I'm coming back from vacation, I might get onto the next five features, but I totally missed the past five, right? Yeah. So how are you design such an MLM interface if you want? In a way that I don't have to like in a way that I'm actually capable of making good use out of that product. Because it almost seems like I missed something. Like I missed a new model. I missed a new feature. Some people even missed core work and they're now jumping right away to that other design thing that just came out yesterday. It's weird. Again, not really sure what's the best way because some of those release will be you know, a must have for some users. They've been waiting for it. Some it's just minor changes in how the model works or tiny, you know, like UI. So I think you can make everyone use your product and all the features. It's impossible. It's just what are essential to the different type of personas. And then how can you make them aware that it exists, but at the same time not forcing them like you don't want just to be okay, go and use that. And sometimes people might use that a lot the first two weeks and then they just give up because in the longer term it's not that useful. So yeah, no, no, I don't have a clearance around that. I think it's just one cool thing is respecting your users. So it's great to have features, but have you, they've been thought in the way that it actually saw real problems that you've heard that has been quantified or is it just based on assumption or because someone internally wanted that? It's also you know at the very beginning that it starts. As a final thought maybe on this episode, we already covered layoffs in the tech industry. And I think whether it's UX designer or researcher or whatever, no one really feels super safe at the moment. But let's maybe go into UX research. Do you have like three tips for our listeners to to not get fired if you want and keep up with with the pace of tech at the moment? I think one honest thing is you can't know everything and I love because I feel like if you're on LinkedIn, you listen to podcasts, you go to events, you have that feeling that you're behind that you are dated, that you should know more and you know we all work a certain amount hours a day and we have to leave. We have to do our whole life. So it's just don't pressure yourself, go at your own pace, try to scale careers, try to stay aware of what's happening, but you don't need to be an expert on everything. And I say the human qualities that you know a lot of people have, collaboration, empathy, all of that people should lean even more. because that's, and the creativity, like the pure creativity, I think as well, like going beyond what exists. Not thinking about constraints, for example, when you're ideating all of those human qualities are for me going to be the next big step. And also how you manage your stakeholders because everyone is very opinionated now, as we've said, everyone can come with their little design with their whole prototype. So being able to navigate as well, those conversation is going to be key. And people should, yeah. I think you can't really develop that overnight, but always looking for opportunities to be, as a UX researcher, you're a bit that kind of middle man where you try to take everyone's use and create the best product possible for the user. So yeah, I would say those are my kind of two, three tips for what to do. Do you have a special way or strategy on how to differentiate, like, actually, a signal from the noise? Because there's just so much to jump onto. And I wonder if you have a certain technique or way of making your decision whether you're going to implement a Claudebot on your Mac Mini over the weekend or you're just, I'm skipping that one and wait for the next trend that is actually having an impact on my work. I think it's a matter of testing a little bit. So, for example, you install the latest experiment it a bit because a lot of people say, oh, this is great, this is not good. And sometimes when you ask them, OK, how have you experimented with it? What two cases? And you realize that they might have used it twice. So I think you have to always test a bit and then see what works out well. In my current company, I don't have access. It's very banking. So data privacy, et cetera, is really strict. So I can only use a certain number of tools. And in my free time, I'm not going to spend all my weekends playing with Claude and Cursor and Ovebo all the time. So it's about being selective as well about what you need and what also brings you joy and fun because it should still be enjoyable. And not just because, oh, I have to test all those five tools and none of them works for me. So you're very much going with your muse, your motivation. Yeah, you're going to test, try, learn, and don't feel like you have to do everything at once because-- Yeah. Yeah, it's just-- I think so, too. Like, sitting on the weekend and going through something like you're not enjoying. You're probably not learning that much anyway. Yeah, no, I think it's like you and I do public speaking. It takes time outside of your working days. So you might as well enjoy-- You want to do a talk that you're going to enjoy sharing and working on rather than, OK, I have seen this way. I try and I have to do it because that's what I am supposed to do as a designer. I think that's the shift. If you want to get to know you better, get to know your work better, what is a good way to reach out? You're nice Ling-Ding is probably the best way. Yeah. So I'm putting your LinkedIn link into the show. Yeah. And that would be a good time to look for that. Yeah, that'd be great. Thank you so much again. Ben, I know it goes in a lot of direction with AI, but hopefully there's some good takeaways for-- There were a bunch of good takeaways. Thank you so much for coming onto the episode. Thank you so much for being here. Thank you guys out there for listening to our episode and see you on the next one. Have a good one. Thanks so for you. Bye. [MUSIC PLAYING]