Bản chép lời đầy đủ của S8E3 Sentient Design for Intelligent Interfaces with Josh Clark and Veronica Kindred

for how it changes the experience of products? 00:02:19.497 --> 00:02:30.354 And so Veronika and I have written a book called Sentient Design, as you said, which kind of codifies the practice that we've developed around designing intelligent interfaces. 00:02:30.354 --> 00:02:32.828 And we'll talk about that some more. 00:02:32.828 --> 00:02:49.112 I will let Veronika introduce herself, but I will say that we are very different generations, that I've been doing this for 30 years, and Veronika is coming into this somewhat new in her career, and we can talk a little bit too about why that's been so important to have those two perspectives. 00:02:49.112 --> 00:02:50.949 But Veronika, over to you. 00:02:51.508 --> 00:02:52.225 Yeah, that's right. 00:02:52.225 --> 00:02:57.895 And um, for listeners who can't see us, I'll politely say that Josh has a few more gray hairs than I do. 00:02:58.372 --> 00:02:59.173 Hey, hey. 00:03:00.609 --> 00:03:01.966 All respect, all respect. 00:03:01.966 --> 00:03:03.801 But yeah, that's right. 00:03:03.801 --> 00:03:14.818 I was in my last year of school when ChatGPT came out when OpenAI dropped ChatGPT in the world, and LLMs really exploded onto the world. 00:03:14.818 --> 00:03:21.683 And I remember my data science professor walking into the classroom very somber one day and just being like, the world has changed. 00:03:21.683 --> 00:03:22.881 And we were all like, what? 00:03:22.881 --> 00:03:24.876 Because like no one had ever really heard of it. 00:03:24.876 --> 00:03:31.342 And then by of course, like the next week, the whole campus was like flooded with LLM-generated content. 00:03:31.342 --> 00:03:32.699 But yeah, that's right. 00:03:32.699 --> 00:03:37.568 So my whole career in design has really been kind of AI native. 00:03:37.568 --> 00:03:49.623 And I've spent the past couple of years working at Big Medium, and I've spent the past couple of years working at Big Medium and really exploring AI as a modality with design. 00:03:50.342 --> 00:04:00.639 Yeah. And you know, I mean, something that I'll say is Veronika is not only my colleague, but she is my daughter, and I'm very proud of her in a lot of different ways. 00:04:00.639 --> 00:04:01.117 Okay. 00:04:01.117 --> 00:04:05.270 So we are we are very literally two generations of design. 00:04:05.908 --> 00:04:06.945 Oh my god, that's so beautiful. 00:04:06.945 --> 00:04:09.500 I'm so glad that you shared that with us. 00:04:09.500 --> 00:04:11.097 So that's fair, Veronika. 00:04:11.097 --> 00:04:16.524 He's definitely many, many gray hairs ahead of me, if I may say so. 00:04:17.165 --> 00:04:18.041 She's allowed to say. 00:04:18.041 --> 00:04:19.079 She's allowed to say. 00:04:19.639 --> 00:04:22.353 All right, and Veronika and Josh, I mean, congratulations. 00:04:22.353 --> 00:04:25.625 This is such an amazing thing to have a father-daughter project. 00:04:26.343 --> 00:04:29.776 It's been a personal and professional highlight for me. 00:04:29.776 --> 00:04:29.937 I can imagine. 00:04:30.336 --> 00:04:31.374 As a parent, I know. 00:04:31.374 --> 00:04:33.449 It's been very fun. 00:04:33.449 --> 00:04:38.478 I think like writing a book is not always the most pleasant task and it is so solitary. 00:04:38.478 --> 00:04:42.870 I think it's really nice to be able to go through it with a close family friend. 00:04:43.827 --> 00:04:44.785 Absolutely. Absolutely. 00:04:45.983 --> 00:04:50.213 I think too, that's just sort of what we're saying is like, as you heard, Veronika doesn't yet have my experience. 00:04:50.213 --> 00:04:57.158 You know, I've been doing this and I've seen a lot of different transitions over the past 30 years, and Veronika's coming to this new. 00:04:57.158 --> 00:05:20.867 So she doesn't yet have my experience, but she's also not burdened by it, which I think is sort of an important thing right now, where what we're seeing is that the possibilities that AI opens for user experience means that we're able to make things that we weren't able to make before and experiences to provide to customers and users that we weren't able to do before. 00:05:20.867 --> 00:05:26.774 And that requires us to go back and say, are those best practices still best practices? 00:05:26.774 --> 00:05:32.920 Just because we've been doing it this way, should we continue doing it this way, or is there a better way? 00:05:32.920 --> 00:05:38.509 And I think having Veronika's fresh eyes has really changed the way that I have worked. 00:05:38.509 --> 00:05:42.821 And I think our practice around this wouldn't be the same. 00:05:42.821 --> 00:05:46.252 And the book certainly wouldn't have been the same without her. 00:05:46.252 --> 00:05:48.168 I couldn't have done this without Veronika. 00:05:49.367 --> 00:05:49.846 Stop it. 00:05:49.846 --> 00:05:54.634 Oh, well, so we are going to have two different perspectives. 00:05:54.634 --> 00:06:05.572 And this is almost like I feel like an inflection point in our field, and where we are not just pivoting, but kind of rethinking and challenging some of the old ways of doing things. 00:06:05.572 --> 00:06:08.285 So what a great way for you two to bring that perspective. 00:06:08.923 --> 00:06:10.440 And you said, what is AI? 00:06:10.440 --> 00:06:22.654 So to give that scope and in your words, describe AI and how it kind of lends itself to this new design material that, you know, rather than a simple tool. 00:06:23.293 --> 00:06:26.966 Yeah, it's a great and surprisingly complex question. 00:06:26.966 --> 00:06:39.019 I mean, we think of AI as this new thing, but it's been with us for gosh, over 75 years now, the better part of a century in one form or another, and it adapts. 00:06:39.019 --> 00:06:49.396 And the really interesting thing, and we see that dialogue happening right now, is at every new step, there's this question of like, is this artificial intelligence? 00:06:49.396 --> 00:06:50.355 Is this it? 00:06:50.355 --> 00:06:51.951 And the conversation now, is

this AGI? 00:06:51.951 --> 00:06:53.387 Is it coming? 00:06:53.387 --> 00:06:56.343 And at every stage, kind of the goalposts keep changing. 00:06:56.343 --> 00:07:01.531 So when we first taught a computer how to play checkers, it's like, well, that's not real intelligence. 00:07:01.531 --> 00:07:02.966 That's just sort of rule following. 00:07:02.966 --> 00:07:04.005 Well, what about chess? 00:07:04.005 --> 00:07:05.043 Oh, no, that's not it. 00:07:05.043 --> 00:07:05.762 Well, what about Go? 00:07:05.762 --> 00:07:06.959 No, that's not it. 00:07:06.959 --> 00:07:11.908 And so the our expectations of what AI means is really different. 00:07:11.908 --> 00:07:18.932 If we had shown the people 75 years ago what large language models can do now, everyone's heads would have exploded. 00:07:18.932 --> 00:07:24.202 I mean, like Veronika was saying, on campus, people's heads did explode when it sort of came out. 00:07:24.202 --> 00:07:29.230 I mean, it's amazing how quickly we say, oh, you know what, that we become used to it. 00:07:29.230 --> 00:07:39.130 I mean, the the things that large language models enable now or the experiences we can have are truly incredibly different from what computing was just five years ago. 00:07:39.130 --> 00:07:51.024 So the way that we're thinking about it is what do we do with this latest era of large language models, things that can do some kinds of reasoning well, but also are incredibly stupid in other places. 00:07:51.584 --> 00:08:08.108 What we talk about with sentient design is it's the practice of creating intelligent interfaces, which are experiences that are aware of context and intent with the agency to adapt to the user in the moment. 00:08:08.108 --> 00:08:11.380 And that can take a whole range of changes. 00:08:11.380 --> 00:08:31.497 But essentially what we're talking about is what happens if we share design decisions in the moment with AI and we teach AI within our guardrails, give it a creative brief and the rules to decide, to make content or behavioral or structural or visual changes in the interface. 00:08:31.497 --> 00:08:37.085 Almost as if a designer was there being like, oh, I can create a custom interface for you in the moment. 00:08:37.085 --> 00:08:57.283 So this can be as wild as websites designing themselves, or small little interventions, or AI having some presence as just another user and a multiplayer experience driving its own cursor around your Figma or Canva board and making comments on your document, just like your colleagues would. 00:08:57.283 --> 00:08:58.080 That's a lot. 00:08:58.080 --> 00:09:03.110 But I mean, that's the idea is like, oh, we can make our experiences more intelligent. 00:09:03.110 --> 00:09:08.857 That's different than how do we use AI to produce stuff, to do the stuff we already do. 00:09:08.857 --> 00:09:12.049 It's about creating genuinely new experiences. 00:09:13.327 --> 00:09:16.999 And thank you for taking the time to kind of explain how you see AI. 00:09:16.999 --> 00:09:22.428 And then what I caught was the experiences that are aware and can adapt. 00:09:22.428 --> 00:09:30.571 Essentially, that's to me, that means that intelligence comes with that definition of sentient design. 00:09:31.448 --> 00:09:31.768 That's right. 00:09:31.768 --> 00:09:36.238 And I don't know, Veronika, it's like we hear a lot of sort of questions, sentient design, eh? 00:09:36.238 --> 00:09:37.835 That seems like an ambitious title. 00:09:38.553 --> 00:09:39.111 Yes, yeah. 00:09:39.111 --> 00:09:47.812 And what we'll say, we don't mean sentient in the in the literal sense, but by sentience, we do mean that awareness of context and intent. 00:09:47.812 --> 00:09:59.947 One of the things that has really taken me and that I've become increasingly interested in in the past year or two is really the combination of probabilistic and deterministic systems. 00:09:59.947 --> 00:10:04.976 So when we talk about things that are aware of context and intent, that's an LLM, right? 00:10:04.976 --> 00:10:08.729 Like you can say something and it will an LLM will understand what you mean. 00:10:08.729 --> 00:10:14.557 You don't have to have like an exact, you don't have to know SQL to be able to manipulate data anymore. 00:10:14.557 --> 00:10:18.947 You should still know data principles, but you know, that's a whole nother thing. 00:10:18.947 --> 00:10:31.799 But combining these algorithms that are probabilistic, like an LLM with deterministic systems, more of like old-fashioned algorithms that can be selectors, unlocks so many possibilities. 00:10:31.799 --> 00:10:52.714 And I think that's really what we're talking about when Josh is talking about kind of systems that can design themselves and systems that are like having a designer there for you in the moment, because you can combine prefab selections that a designer has approved and that makes sense for the customer experience and for the journey, and that's all been made and decided. 00:10:52.714 --> 00:10:59.740 But an LLM can come in and probabilistically choose the right one for the moment. 00:10:59.740 --> 00:11:13.868 So that's the combination of things that gets really interesting because there's no customer journey, there's no happy path, but you still have these reliable pieces, Lego pieces that you can like build together for an experience. 00:11:14.745 --> 00:11:19.297 I'm thinking doing something without a customer involved is a pretty bold thing to do, right? 00:11:19.936 --> 00:11:24.644 So to be clear, right, it has to begin with the customer's needs and of course the business needs too. 00:11:24.644 --> 00:11:26.082 And how do you marry those? 00:11:26.082 --> 00:11:39.971 And I think the the way that we've managed that traditionally is by defining some sort of course personas or segments and then designing static paths for those. 00:11:39.971 --> 00:11:41.489 It's the best we could do, right? 00:11:41.489 --> 00:11:46.437 It's like in truth, there's a million

different scenarios that users bring to things. 00:11:46.437 --> 00:11:50.350 And we can't historically design a million scenarios. 00:11:50.350 --> 00:11:57.932 So maybe we design seven different dashboards for these different scenarios, or people individually create things. 00:11:57.932 --> 00:12:02.245 And then over time, it's like, oh, we've got a thousand dashboards that now people have to navigate. 00:12:02.245 --> 00:12:10.067 And so the idea here is not to say, oh, we don't care about what the user thinks, or that we're not designing around those user needs. 00:12:10.067 --> 00:12:25.155 But rather we're saying, what could we do if AI could help interpret that need in the moment and bring that experience to the user instead of make them climb through some complex navigation? 00:12:25.155 --> 00:12:45.990 So this can sound somewhat abstract, but really what we're doing, and this is what Veronika is talking about about sort of an LLM working within a deterministic system, is that we still, as designers, create the structure and the environment that both humans and AI will work within. 00:12:45.990 --> 00:12:51.977 So I don't know, as a very more concrete example, Veronika, do you want to talk about like Salesforce or Cisco? 00:12:51.977 --> 00:12:56.129 So this sounds like, wow, this is kind of a wild, fluid world. 00:12:56.129 --> 00:12:57.565 Who's doing this? 00:12:57.565 --> 00:13:04.030 And it turns out even in the very kind of conservative, buttoned-up world of enterprise software, we're seeing it. 00:13:04.990 --> 00:13:08.582 right. The Salesforce example is something called Salesforce Generative Canvas. 00:13:08.582 --> 00:13:13.291 And it takes the context of a user. 00:13:15.765 --> 00:13:20.796 at a user's calendar and it sees that they have a meeting coming up with a client. 00:13:20.796 --> 00:13:37.080 This and we'll say this is a sales team member, and they see that they have a meeting coming up with the client, and they assemble their dashboard such that there are notes from the last time they had a meeting, the most up-to-date stats on that customer in the Salesforce database. 00:13:37.080 --> 00:13:44.745 And so it's assembling all of these components in the moment way, using context from the user's calendar. 00:13:44.745 --> 00:13:49.293 You can also ask it a direct question and it will and it will populate the canvas. 00:13:49.293 --> 00:13:54.163 But what's great here is that the data is not being LLM generated. 00:13:54.163 --> 00:14:05.898 The content of the components is not being made in the moment, but it's taking the data that already exists in the Salesforce system and bringing it to the user so that they don't have to navigate throughout. 00:14:05.898 --> 00:14:17.714 And you can definitely imagine this being really useful in like a client meeting where you're trying to stay focused, maintain good eye contact, even over Zoom, but you still have kind of the things you need coming to the service. 00:14:17.714 -->

00:14:20.506 So you can just quickly glance and keep talking. 00:14:21.306 --> 00:14:22.981 It's not like making up the design, right? 00:14:22.981 --> 00:14:30.086 That's got this little sort of scoped design system of sort of different widgets that are tuned to present certain kinds of information. 00:14:30.086 --> 00:14:32.402 And it's got certain sources of information. 00:14:32.402 --> 00:14:36.312 And it knows kind of like, oh, here's this sort of dashboard archetype. 00:14:36.312 --> 00:14:38.707 So there's rules involved, right? 00:14:38.707 --> 00:14:43.418 And AI is sort of saying, oh, here's the context, here's the kind of data that I should do. 00:14:43.418 --> 00:14:44.296 Let me fetch that. 00:14:44.296 --> 00:14:48.368 Oh, and here, because it's this kind of data, here are the right components to display. 00:14:48.368 --> 00:14:58.186 And so it's making sort of simple decisions within a fixed system, but that can give this a huge variety of results to the user. 00:14:58.903 --> 00:15:00.899 Now, to a large extent, it makes sense to me. 00:15:00.899 --> 00:15:07.686 I'm thinking right now, I'm trying to build a daily digest where I get to see what my day is going to be looking like. 00:15:07.686 --> 00:15:16.625 And maybe if an LLM can create a dashboard for me as the day changes, as my priorities change, that's a very positive user experience. 00:15:16.625 --> 00:15:18.222 Would that be something about it? 00:15:18.462 --> 00:15:19.660 Yeah, that's exactly it. 00:15:19.660 --> 00:15:23.013 And right, like you think about the different types of days that we have. 00:15:23.013 --> 00:15:25.886 Oh, this is a thinking day where I'm creating and making. 00:15:25.886 --> 00:15:29.159 What are the materials or context that I need for that? 00:15:29.159 --> 00:15:30.437 Or this is a meeting day. 00:15:30.437 --> 00:15:34.748 Like, remind me who these people are, what we decided the last time, what's next. 00:15:34.748 --> 00:15:41.452 Like those are very different kinds of experiences that you want to have different information. 00:15:41.452 --> 00:15:44.248 How can AI help to make that? 00:15:44.248 --> 00:15:56.539 Like we've been talking about adaptive interfaces and personalization for decades, but we finally have a technology that can actually help to do that on the fly at the presentation level, right? 00:15:56.539 --> 00:15:59.254 Like it's not about AI making up facts. 00:15:59.254 --> 00:16:04.604 It's about making, oh, really sort of simple decisions from available design material. 00:16:04.604 --> 00:16:12.427 And so as designers, we create that environment and then we teach AI to work within that environment. 00:16:12.427 --> 00:16:18.493 And that sort of behavior design, and this is like what technical communicators are great at. 00:16:18.493 --> 00:16:47.392 Like it's like, oh, let me actually explain how you do this thing. That skill becomes incredibly important at this sentient design layer because you're teaching do's and

don'ts, you're teaching behavior, you're kind of giving, in a way, a little a junior designer the brief for how to respond to the user and what to do, what happens when it's a confusing situation, what happens when the data is ambiguous? 00:16:47.392 --> 00:16:49.547 What do you do in these situations? 00:16:49.547 --> 00:16:59.605 And so, in a sense, technical communicators have a new audience, that it's not just the people on the company side and the people on the customer side. 00:16:59.605 --> 00:17:05.751 It's this enabling collaborative layer that AI provides at generating these experiences. 00:17:07.108 --> 00:17:08.467 Veronika, would you like to add something? 00:17:08.467 --> 00:17:09.184 Thanks, Josh. 00:17:10.462 --> 00:17:13.815 Yeah. Maybe I'll just talk about another example, if that's all right. 00:17:13.815 --> 00:17:13.973 right. 00:17:13.973 --> 00:17:15.571 Go for it. 00:17:15.571 --> 00:17:15.970 Yeah. 00:17:15.970 --> 00:17:17.646 So we talked about the Salesforce example. 00:17:17.646 --> 00:17:22.356 The other example that Josh mentioned is the Cisco dashboard. 00:17:22.356 --> 00:17:28.903 And the Cisco dashboard is similar to the Salesforce canvas in that it's assembling things on the fly. 00:17:28.903 --> 00:17:38.083 But rather, if you can picture it visually, on the left hand side you have a kind of traditional LLM chat, and on the right hand side, you have like a big dashboard. 00:17:38.083 --> 00:17:47.104 And as you ask questions of the chat, it starts to populate the dashboard in in in response to your question. 00:17:47.104 --> 00:17:56.124 You can also ask specific questions about what's already on the dashboard, and it might show you a Viz in chat, which you can then drag onto the dashboard. 00:17:56.124 --> 00:18:00.993 So in this situation, you're really able to assemble dashboards. 00:18:00.993 --> 00:18:03.468 This has a heavier hand, right? 00:18:03.468 --> 00:18:09.057 You're assembling dashboards according to what you need and according to your questions rather than kind of imbued context. 00:18:09.057 --> 00:18:12.009 And then you could see how you can push this further. 00:18:12.009 --> 00:18:13.924 Like, oh, can I share this dashboard? 00:18:13.924 --> 00:18:15.921 Can I invite people onto my dashboard? 00:18:15.921 --> 00:18:19.035 Can I save it for later and come back to it? 00:18:19.035 --> 00:18:25.420 So there's all these possibilities when you start thinking of these dashboards that really bring content to you. 00:18:26.138 --> 00:18:27.176 This is just one of the examples. 00:18:27.176 --> 00:18:29.571 You know, I know we're being sort of very dashboard focused. 00:18:29.571 --> 00:18:32.445 It tends to be an example that's kind of easy to visualize. 00:18:32.445 --> 00:18:33.483 It's like, all right. 00:18:33.483 --> 00:18:36.038 It's like these elements that are being assembled on the fly. 00:18:36.038 --> 00:18:42.265 That's a pattern that we call bespoke UI, which is just like a custom interface that's being generated for users. 00:18:42.265 --> 00:18:48.891 But we have 14 other experiences, including, you know, how do you work with agents right now? 00:18:48.891 --> 00:18:54.078 Sort of this mainstream idea of agents, it's a very text-based, almost console-based experience. 00:18:54.078 --> 00:19:00.705 What happens when you have that, those actions and actors in the interface itself? 00:19:00.705 --> 00:19:12.279 So that they are in the same way, if you think about an agent as a multiplayer experience, like in other collaborative environments, how should they show up as collaborators in the experience? 00:19:12.279 --> 00:19:19.625 When you're editing a document, we know how one of our colleagues would edit in the comment and markup a document. 00:19:19.625 --> 00:19:23.296 What happens when AI shows up in that interaction? 00:19:23.296 --> 00:19:26.410 Or in a Canvas experience like Figma? 00:19:26.410 --> 00:19:31.997 What if an agent has its own cursor to do tasks for you or add comments? 00:19:31.997 --> 00:19:49.082 And so it's it's this thing of really about weaving intelligence into the interface so that the interface itself becomes a collaborator with the user or is responding in real time, as Veronika's mentioned in the Salesforce and and Cisco examples. 00:19:49.719 --> 00:19:50.758 And I'll say what they're not plugs. 00:19:50.758 --> 00:19:55.147 We're not no one's paying us to talk about them. 00:19:56.105 --> 00:19:58.181 Salesforce, Salesforce makes enough money. 00:19:58.660 --> 00:19:59.857 Yeah, no disclosures here. 00:20:00.819 --> 00:20:04.980 Do you know how this is being used broadly, or how it will become applicable? 00:20:05.759 --> 00:20:11.366 The episode is sponsored by LavaCon, the content strategy conference taking place this October in Charlotte, North Carolina. 00:20:11.366 --> 00:20:18.096 With more than 70 sessions and workshops featuring speakers from companies like Salesforce, TikTok, T-Mobile, and Amazon. 00:20:18.096 --> 00:20:25.945 LavaCon brings together content strategists, documentation leaders, and content professionals to share practical ideas you can apply right away. 00:20:25.945 --> 00:20:32.994 And yes, it's known as the Fun Conference, too, with networking events, therapy dogs, comfort llamas, storytelling, and karaoke. 00:20:32.994 --> 00:20:37.240 Use discount code ITC26 to save $200 on registration. 00:20:37.920 --> 00:20:41.440 But what I want to ask you, Veronika, is you know where I'm looking from. 00:20:41.440 --> 00:20:46.159 I am a Gen Xer and I have kids who are probably just a little younger to you. 00:20:46.159 --> 00:20:51.838 I look at AI with I used to look at AI with fear, not as much anymore. 00:20:51.838 --> 00:20:54.398 But how are you looking at your future? 00:20:54.398 --> 00:20:57.519 Like I want to bring I want to understand your take on that. 00:20:58.157 --> 00:20:59.759 Yeah, thanks so much for that question. 00:20:59.759 --> 00:21:02.397 It's

definitely something I've thought a lot about. 00:21:02.397 --> 00:21:07.278 And I think the fear that you used to feel, I think a lot of people still feel that. 00:21:07.278 --> 00:21:11.116 And there's like a lot of anxiety around AI and AI adoption. 00:21:11.116 --> 00:21:13.836 And it makes so much sense. 00:21:13.836 --> 00:21:25.037 Like AI is having a pretty tangible impact on the workforce, and people are worried about jobs or they've lost their jobs, partly in thanks to AI automation. 00:21:25.037 --> 00:21:26.477 And that's so scary. 00:21:26.477 --> 00:21:28.717 Like there's so many reasons to be fearful. 00:21:28.717 --> 00:21:31.997 And there's also like the other impacts of AI. 00:21:31.997 --> 00:21:42.715 Like, we don't know yet the social and economic impacts of this technology, and it's kind of being allowed to run rampant without great regulation or legislation surrounding it. 00:21:42.715 --> 00:21:47.036 So I think that's where a lot of my anxiety around it still comes in. 00:21:47.036 --> 00:21:52.315 It's like, oh wow, I really wish we were addressing this more head-on at an institutional level. 00:21:52.315 --> 00:22:08.633 That seems like a whole separate podcast, but I think what gives me hope is I think thus far the focus on AI has been on efficiency and has been on making things faster or cheaper. 00:22:08.633 --> 00:22:11.753 And that is just such a race to the bottom. 00:22:11.753 --> 00:22:15.353 Like, you know, who can make this the fastest and the cheapest? 00:22:15.353 --> 00:22:18.313 Has what about quality, you know? 00:22:18.313 --> 00:22:22.313 And so I think so much focus thus far has been on tooling. 00:22:22.313 --> 00:22:31.271 And even in like the design industry, a lot of our focus has been on like the tooling of AI, like, oh, how can I make my designs faster? 00:22:31.271 --> 00:22:33.991 Or how can I expedite this whole process? 00:22:33.991 --> 00:22:35.111 Like all of this. 00:22:35.111 --> 00:22:43.191 And what there's not been much focus on, and what Josh and I have tried really hard to focus on, is making new kinds of things. 00:22:43.191 --> 00:22:45.990 You know, what new products can I make? 00:22:45.990 --> 00:22:48.550 What new experiences the This enable? 00:22:48.550 --> 00:22:56.630 And I think that has yet to hit mainstream or to hit other industries, not just our book specifically, though you know, I hope it does. 00:22:56.630 --> 00:23:02.230 But just that idea of like, okay, this technology does mean we can do things faster. 00:23:02.230 --> 00:23:06.230 It does mean we don't have to write long emails anymore. 00:23:06.230 --> 00:23:13.907 And that's where a lot of people stop thinking about it is like, oh, how can this help me do what I already did with lesser effort? 00:23:13.907 --> 00:23:21.428 And there's not been a ton of innovation or imagination around what new types of things it can do. 00:23:21.428 --> 00:23:24.146 So I think that's where I get very hopeful and excited. 00:23:25.028 --> 00:23:26.948 You and I are generational peers, Zohra. 00:23:26.948 --> 00:23:35.666 And so there's like a lot of this thing that is, oh man, all right, this does this threaten my decades of experience and hard-won knowledge? 00:23:35.666 --> 00:23:37.425 Is this just gonna sort of take over? 00:23:37.425 --> 00:23:48.865 I will say that I have never been so excited as a designer than I am right now in this career for exactly the reasons that Veronika is mentioning about sort of the upsides of this. 00:23:48.865 --> 00:23:59.506 If we shift our thinking from how do we save money, how do we cut, how do we do what we have always done, but faster, that's fine. 00:23:59.506 --> 00:24:02.066 There's like there's value to that for sure. 00:24:02.066 --> 00:24:03.903 I like productivity, I like efficiency. 00:24:03.903 --> 00:24:07.344 But what's missing is how do we create new value? 00:24:07.344 --> 00:24:19.182 How do we create something that we couldn't do before that can create a tremendously valuable and meaningful experience for both customer and business? 00:24:19.182 --> 00:24:20.784 Like that's exciting. 00:24:20.784 --> 00:24:23.903 We have new material to work with. 00:24:23.903 --> 00:24:35.022 And that's the shift that Veronika is talking about is going from a tool for production to a material for product. 00:24:35.022 --> 00:24:38.702 And product ultimately is what we're all here for. 00:24:38.702 --> 00:24:40.782 Production is just a means to an end. 00:24:40.782 --> 00:24:46.701 What the only thing that matters is what we put in front of the customer and what happens next. 00:24:46.701 --> 00:24:51.901 And we have the opportunity to invent meaningful new ways to do that. 00:24:51.901 --> 00:25:04.060 And when you think about that, right, my job isn't just about the production, it's about imagining what's possible and now working with this new to the world material to make those things happen. 00:25:04.060 --> 00:25:05.500 That is exciting. 00:25:05.500 --> 00:25:08.540 And that is design and content work. 00:25:08.540 --> 00:25:14.378 You know, there's the the old saying in product management of good, fast, cheap, pick two. 00:25:14.378 --> 00:25:21.179 If fast and cheap are just available to everybody, then the only thing that matters is good, right? 00:25:21.179 --> 00:25:23.178 Is what shall we make? 00:25:23.178 --> 00:25:24.459 What shall we imagine? 00:25:24.459 --> 00:25:29.097 And that we have a whole new world to invent right now. 00:25:29.097 --> 00:25:42.298 With sentient design, we're trying to establish what are the foundations that are emerging now, what are sort of some of the principles that we can use to invent these new kinds of experiences, but there's still so much invention ahead. 00:25:42.298 --> 00:25:53.257 And I think that's gonna take a generation of work to figure out in the same way that the web itself and mobile took generations to figure out. 00:25:53.257 --> 00:25:56.135 And that's the

work, Veronika, for you ahead. 00:25:57.415 --> 00:26:02.615 is the one thing that you can say if you're kind of worried, looking through the lens of where you are sitting? 00:26:03.414 --> 00:26:09.496 I mean, jobs are changing, and some of the tasks that we associate with our role will go away. 00:26:09.496 --> 00:26:12.695 And some jobs may go away too, some roles may go away. 00:26:12.695 --> 00:26:18.454 And you know, I mean, I don't think it's anything super new to say that production is at risk, right? 00:26:18.454 --> 00:26:33.973 The stuff that it used to take, you know, that especially around code, some aspects of writing, although that really depends on the context, are at risk that AI can just crank this stuff out or at least make an incredibly credible first draft. 00:26:33.973 --> 00:26:47.892 And so the shift has to be around we are more than just generators of code or generators of word, that we think about what's important, why are we creating this thing? 00:26:47.892 --> 00:27:03.971 And so, in some ways, while a lot is changing in terms of the production of our work, all of those kinds of efficiency things that Veronika was talking about, or what we can just ask Claude or ChatGPT or Gemini or Copilot to just please just do this thing and it happens. 00:27:03.971 --> 00:27:06.611 That is at risk for sure. 00:27:06.611 --> 00:27:08.369 And that work is changing. 00:27:08.369 --> 00:27:17.010 But I think it also lets us use these things as material to create new things, like we were talking about. 00:27:17.010 --> 00:27:28.690 And I think that that's especially important at the at the product level of you know, what can we get show to customers, not just how do we get to it faster, but what can we show to them that's new? 00:27:28.690 --> 00:27:32.368 But also like the foundations of our work don't change. 00:27:32.368 --> 00:27:43.647 So, especially for Zohra, for folks like you and I who have been at this for a while, I want to give some reassurance that it's still about the fundamental questions of what problems are we trying to solve? 00:27:43.647 --> 00:27:45.567 What frictions can we overcome? 00:27:45.567 --> 00:27:50.688 And now we have a new material to work with to do that. 00:27:50.688 --> 00:27:54.768 Like we've just like when the web came around, it's like, oh, this is new. 00:27:54.768 --> 00:27:59.886 This is a new vocabulary to learn as a creator and as a consumer. 00:27:59.886 --> 00:28:07.165 Same with mobile, you know, and so every 10 or 15 years we get one of these things that are both crisis and opportunity. 00:28:07.165 --> 00:28:23.246 And so I think that the way to think about it and to sort of future-proof our jobs or adapt our roles is how do I work with this new material, not as a tool, but as a something that can deliver outcomes in new ways. 00:28:23.246 --> 00:28:44.363 And I think an exciting and challenging thing is that this does change the process of our work and the kind of collaboration that we

can have, that this idea of working together to help AI make good decisions really calls for more collaboration. 00:28:44.363 --> 00:28:51.163 And the cool thing is that the work surface and workspace can be the same. 00:28:51.163 --> 00:28:55.801 And by the way, this is like stuff that technical communicators are can be really great at. 00:28:55.801 --> 00:29:14.922 That if we think of, you know, really what we're talking about is giving systems the context and instructions that they need to make good decisions, whether they're writing technical documentation or we're teaching them the behavior of creating one of these bespoke interfaces that Veronika was talking about. 00:29:14.922 --> 00:29:18.601 product, technical communicators, 00:29:18.601 --> 00:29:20.520 designers, developers. 00:29:20.520 --> 00:29:24.839 And traditionally, that's been a waterfall step, right? 00:29:24.839 --> 00:29:28.441 You know, despite all the talk about Agile, we're not working together on this stuff. 00:29:28.441 --> 00:29:29.799 But now we can. 00:29:29.799 --> 00:29:43.000 We can, through all of the phases of a product development, we can all contribute our discipline to this common workspace, this common system prompt, the context that goes in it together. 00:29:43.000 --> 00:29:51.317 And one of the things that Veronika and I are experiencing a lot is that all of our teams are working together in the same common repo. 00:29:51.317 --> 00:30:01.478 And even if we aren't coding, we are contributing the documentation that AI needs to make good decisions in the moment. 00:30:01.478 --> 00:30:02.917 And it's been really powerful. 00:30:02.917 --> 00:30:09.396 The risk is that we do things, is that AI makes me think, oh, I can do all of this on my own without my colleagues. 00:30:09.396 --> 00:30:11.958 I'm a designer, but now I can deliver code. 00:30:11.958 --> 00:30:12.758 I don't need developers. 00:30:12.758 --> 00:30:15.635 I'm a developer, but now I AI can do the design. 00:30:15.635 --> 00:30:16.676 I don't need designers. 00:30:16.676 --> 00:30:20.355 That's a risk because then everyone runs in different directions. 00:30:20.355 --> 00:30:27.795 The real opportunity, and this is another area to think about for career, is how do we collaborate better? 00:30:27.795 --> 00:30:38.675 How do we bring people together to make decisions together quickly and iteratively in the same workspace that is product focused, customer focused? 00:30:38.675 --> 00:30:40.275 Those are new skills. 00:30:40.275 --> 00:30:46.194 I think there's a lot of talk about, oh, you can be a 10x designer, you can be a 10x technical communicator. 00:30:46.194 --> 00:30:48.515 What about your team? 00:30:48.515 --> 00:30:51.234 Like this is, you know, how do we make our team awesome? 00:30:51.234 --> 00:30:54.034 And more to the point, how do we make our customer awesome? 00:30:54.753 --> 00:31:02.353 Well, I'd love to kind of pick up on something Josh was talking about, which is just this idea of like a shared workspace within a team. 00:31:02.353 --> 00:31:16.032 I think with the I think with AI coming into the mix, there's been this kind of flattening of skills where, like Josh was saying, like now when I'm designing, most of the time I'm designing in Claude Code. 00:31:16.032 --> 00:31:25.070 Like I use Figma probably 80% less than I did last year, which is just a crazy difference in process for one year. 00:31:25.070 --> 00:31:34.191 And I think what's really interesting is now when I'm in Figma, more often I'm in a FigJam board rather than a Figma design file. 00:31:34.191 --> 00:31:39.390 And more often I'm in that FigJam board specifically to collaborate with people from different disciplines. 00:31:39.390 --> 00:31:47.470 So having like a product person and a developer all in the fig jam board working on the same thing and all trying to get our ideas out. 00:31:47.470 --> 00:31:50.348 So there's less technical barriers. 00:31:50.348 --> 00:32:00.269 And so just to kind of pull a thread that Josh was picking out as well, it becomes even more important to really have your discipline's perspective. 00:32:00.269 --> 00:32:20.748 Like I still need to, even though I'm working on code technically, what you know, and you know, that gets sent to the developer who's like making sure it's healthy, just to catch any Claude-isms or anything. But it becomes so important for me to have that design perspective and to really hone down on like the main thing a designer is there for, which is like, how am I looking out for the customer? 00:32:20.748 --> 00:32:23.866 Like, how am I making sure the customer is getting what they need from this experience? 00:32:23.866 --> 00:32:29.385 And and the techniques, which you know, last year was like, how well can you use Figma? 00:32:29.385 --> 00:32:37.467 And don't get me wrong, like I can use Figma, but it's becoming less important as the perspective that you're bringing and the principles that you're bringing to it. 00:32:37.467 --> 00:32:38.905 So I think that's been really exciting. 00:32:38.905 --> 00:32:41.705 It's like a strange distillation of the craft. 00:32:42.425 --> 00:32:48.266 I do think there's sort of an evolution of how we adopt new technology that that I think Veronika is calling out here. 00:32:48.266 --> 00:32:50.425 Is first we start with, oh, we have these tools. 00:32:50.425 --> 00:32:53.064 How do we fit this into our existing process? 00:32:53.064 --> 00:32:55.944 How do we do what we already do faster? 00:32:55.944 --> 00:32:58.344 And then that has some knock-on effects. 00:32:58.344 --> 00:33:00.265 It's like, oh, wait, I'm I'm working differently. 00:33:00.265 --> 00:33:02.265 Now we as a team have to work differently. 00:33:02.265 --> 00:33:05.303 How do we change process and maybe organization? 00:33:05.303 --> 00:33:08.983 Those things can be hard, especially in large organizations. 00:33:08.983 --> 00:33:11.223 That's not nothing, right? 00:33:11.223 -->

00:33:17.943 And I think that most companies are there right now in the between those first and two stages of changing our tooling. 00:33:17.943 --> 00:33:19.702 Now, how do we change our process? 00:33:19.702 --> 00:33:23.222 And then that third bit is like, how do we change our product? 00:33:23.222 --> 00:33:36.580 And that's the bit that that Veronika and I are especially fascinated with and that we help our clients work with is like, what is the new thing that instead of just making the old thing faster, how can we make the new thing? 00:33:36.580 --> 00:33:37.780 And what are the opportunities? 00:33:37.780 --> 00:33:39.221 Does that open? 00:33:39.221 --> 00:33:46.260 And I think that's the most exciting thing for the next generation of product design and product organizations. 00:33:46.260 --> 00:33:52.020 Because from there, what happens is sometimes those things open up entirely new lines of business. 00:33:52.020 --> 00:33:57.461 And in some small cases, might even disrupt a whole industry and sort of reshape an industry. 00:33:57.461 --> 00:34:05.538 Most companies aren't in the disruption business, you know, that we're sort of like looking at, you know, how do we do, how do we provide our services now better? 00:34:05.538 --> 00:34:11.378 And I think that is like an especially exciting thing about that product invention phase. 00:34:11.378 --> 00:34:12.498 So it's okay. 00:34:12.498 --> 00:34:15.137 It's okay if you recognize it's like, oh, you know what, we're still doing tooling. 00:34:15.137 --> 00:34:16.739 We're not saying you're doing the wrong thing. 00:34:16.739 --> 00:34:18.259 That is where things begin. 00:34:18.259 --> 00:34:24.097 And learning to use those tools often say, oh, what if we did that in our experience, in our product? 00:34:24.097 --> 00:34:26.257 How did or what new product would we create? 00:34:26.978 --> 00:34:31.777 I think we need to step away and kind of really think about where can I, what can I do differently? 00:34:33.137 --> 00:34:45.295 Yeah. I mean, I think one of the things that's interesting too, like Veronika was saying, there's so many reasons to be skeptical or even fearful of this change and what it means for us individually and as a field. 00:34:45.295 --> 00:34:47.697 There are plenty of reasons for that. 00:34:47.697 --> 00:35:02.416 I guess I would say that skepticism and fear should be more reason to engage with this, not less, and not least because the technology changes so quickly, that the capability changes. 00:35:02.416 --> 00:35:05.775 And so thus, so it's the possibility and creative material. 00:35:05.775 --> 00:35:25.934 And so I think one thing is if, for example, you're someone who hasn't tried to use AI to do a task in say six months, because it was sort of meh then, you may be very surprised at how well it can do that thing now, either in a customer-facing way or how it supports your own work. 00:35:25.934 --> 00:35:37.692 And so I think one of these things is if you think of AI as your enemy, well, then know your enemy because soon you might be able to like understand it better, make it a friend. 00:35:37.692 --> 00:35:51.371 I don't want to sort of like over-personalize uh what AI is here, but I do think there is this thing of like understanding the material will help you use it in ways that serve not only yourself, but the customer and the business. 00:35:52.012 --> 00:35:53.851 Veronika, would you like to add something? 00:35:54.570 --> 00:36:03.530 Yeah, I not to be a downer, but I just want to acknowledge, I think there are some barriers that lead to people not adopting new tools. 00:36:03.530 --> 00:36:14.329 You know, if you are in a huge enterprise, you know, there's very real like firewalls and you can access this and that type of thing, as well as yeah, that that intimidation factor. 00:36:14.329 --> 00:36:31.849 So for anyone who maybe feels like they face those kinds of obstacles, I would just want to give a message of encouragement to just try and even even for a month, you know, just to really like push it as far as you can and use it in in different ways, not just for work. 00:36:31.849 --> 00:36:36.489 Zohra, you're talking about, oh, like maybe I can assemble this dashboard to help me get through my day. 00:36:36.489 --> 00:36:37.686 And it's like, yes, totally. 00:36:37.686 --> 00:36:41.927 And Josh, you were talking about, you know, is it a thinking day or is it a meeting day? 00:36:41.927 --> 00:36:46.567 And I think it's also like, well, are you responsible for picking kids up from school? 00:36:46.567 --> 00:36:48.327 Do you have to go grocery shopping later? 00:36:48.327 --> 00:36:51.367 Can it give you the recipe when you need the recipe and all these things? 00:36:51.367 --> 00:36:59.447 Like, because I think using it in non-work circumstances is really been a place where like my imagination around it grows. 00:36:59.447 --> 00:37:07.125 And I'm like, or I discover new tooling, or I'm like, I didn't know Claude could do this because I'm like doing it out of a context that I like think that I already know. 00:37:07.605 --> 00:37:11.445 I mean, that's getting at that perspective shift from tool to material. 00:37:11.445 --> 00:37:29.523 Like a lot of things, if you can shift your thinking as a technical communicator, as a designer, as a product leader, from how do I just use AI over here in this specific role of making the thing that I need to make to how do I incorporate it into the thing that I make? 00:37:29.523 --> 00:37:38.242 That is a big unlock, is what we've discovered of creative possibility for the individual and product for the end user. 00:37:38.963 --> 00:37:42.563 This is definitely, I think, stepping outside of what you're doing, Veronika. 00:37:42.563 --> 00:37:45.041 I want to go back to your thought, what that you shared. 00:37:45.041 --> 00:37:49.842 Stepping away from what you do day to day and thinking creatively can actually be fun. 00:37:49.842 --> 00:37:53.762

So interact with that's that's such a fun way of thinking. 00:37:55.282 --> 00:37:56.561 There is this thing too, right? 00:37:56.561 --> 00:38:07.121 It's like in projects now, we reserve some time for what we call mad science to experiment and play around the subject instead of rushing to solution. 00:38:07.121 --> 00:38:13.920 Because right now there is opportunity to discover new ways to accomplish something or to present something. 00:38:13.920 --> 00:38:20.079 And so I think it's really important that kind of play and experimentation that both of you are talking about. 00:38:20.481 --> 00:38:26.320 Yeah, I love the phrase on that side because oh, that that totally vibes with me, I think. 00:38:26.639 --> 00:38:32.719 Not that I'm a scientist, but uh, feel like one, yeah. 00:38:33.199 --> 00:38:35.199 Yeah, I know, or why not feel like one, right? 00:38:35.199 --> 00:38:37.518 Uh and I think AI allows me to think that. 00:38:37.518 --> 00:38:38.318 So why not? 00:38:38.878 --> 00:38:40.157 Absolutely. You're like, I can code. 00:38:40.157 --> 00:38:42.077 I'm like super I'm so STEM. 00:38:43.358 --> 00:38:46.077 You introduced the ideal of defensive design. 00:38:46.077 --> 00:38:48.077 What do you mean by that? 00:38:48.637 --> 00:38:51.998 We don't want to kind of come off as boosters and that AI can do everything. 00:38:51.998 --> 00:39:03.996 All of us have had the experience working with AI where it's like that's not right, or it's just made that up, or terrible assumptions, or we've all had experiences where things go sideways. 00:39:03.996 --> 00:39:18.476 And so, you know, I think Veronika, if you want to talk a little bit about this, we've had this kind of strategy around defensive design, both as a mindset and specific patterns that you can use to sort of promote safety and trust. 00:39:19.195 --> 00:39:19.594 Yeah, absolutely. 00:39:19.594 --> 00:39:25.114 Yeah, defensive design is is really the practice of designing in guardrails. 00:39:25.114 --> 00:39:34.713 So rather than just right now, you know, you're using an LLM and there's like a bit of legalese below a chat bot, it's like answers might not be right. 00:39:34.713 --> 00:39:35.594 Double check. 00:39:35.594 --> 00:39:42.152 Or there's, you know, you see a new feature on a website and there are some purple sparkles on it. 00:39:42.152 --> 00:39:45.273 And those purple sparkles are symbols of AI. 00:39:45.273 --> 00:39:49.512 So you know you're using AI, like we're all very well adjusted to that symbol now. 00:39:49.512 --> 00:39:58.632 But it's kind of giving you the message of like this might not work, like this is for fun, this isn't you know central to your task, or it can't exactly be relied upon. 00:39:59.192 --> 00:39:59.672 Do not trust. 00:40:00.632 --> 00:40:01.592 Do not trust, yeah. 00:40:01.592 --> 00:40:10.710 So rather than designing in guardrails, we're getting used to these kinds of like caution or warning signs. 00:40:10.710 --> 00:40:16.391 So, which really kind of go to undermine the product because if it doesn't work, why are you putting it on your website? 00:40:16.391 --> 00:40:21.831 That said, of course, AI hallucinates, it can't be trusted 100% of the time. 00:40:21.831 --> 00:40:29.510 It does a remarkably good job of giving you things that look like answers, and in a very miraculous way, it's often right. 00:40:29.510 --> 00:40:31.271 Like it is often the correct answer. 00:40:31.271 --> 00:40:37.028 But there is, you know, that 10 or 5 or 1% of the time where it's wrong. 00:40:37.028 --> 00:40:41.750 And in a creative context, that can be fine, fun, whatever. 00:40:41.750 --> 00:40:46.068 That picture now has six fingers on somebody's hand, like that's funny. 00:40:46.068 --> 00:40:49.108 But in other situations, that's a real problem. 00:40:49.108 --> 00:40:58.389 If you're in healthcare or airlines or any kind of safety regulation that is super, super high stakes, you can't be wrong like that. 00:40:58.389 --> 00:41:07.268 So, how can you build in signals that encourage productive humility within the system? 00:41:07.268 --> 00:41:15.907 So the system is not being too arrogant about its own abilities, as well as some healthy skepticism on the part of the user without just like adding those sparkles in. 00:41:15.907 --> 00:41:18.786 And there's a number of different ways to do this. 00:41:18.786 --> 00:41:23.347 You know, there are things to be said for the way something is worded. 00:41:23.347 --> 00:41:37.585 There was an interesting experiment done that said if you present percentages to people, like if you're like this is 90% likely to be an elephant and it's a picture of a giraffe or something like that, but that doesn't mean much to people. 00:41:37.585 --> 00:41:44.784 There's a similar thing, maybe this is a better example on on Netflix where it like has a movie and it's like you are 74% likely to like this movie. 00:41:44.784 --> 00:41:46.465 And it's like, what does that mean? 00:41:46.465 --> 00:41:49.985 Like, what does 74% likely mean? 00:41:49.985 --> 00:42:01.182 But if you're wording something and saying, like, um, you're like probably gonna like this movie, there's a lot more information being given, also in my tone, right? 00:42:01.182 --> 00:42:08.702 Like, you're probably gonna like, you know, it's you know, we're able to understand that it's like, oh, I don't know, I might not, you know. 00:42:09.021 --> 00:42:13.983 Right. But saying it's saying I'm 10% confident is different from I really have no idea. 00:42:13.983 --> 00:42:19.581 So I'm making this up a little bit, but here's a guess based on very slim data. 00:42:19.581 --> 00:42:20.942 Like that's really different. 00:42:20.942 --> 00:42:28.141 And it's just like, oh, I'm listening to you now and I'm understanding so so there's a presentation aspect, not just a clinical. 00:42:29.181 --> 00:42:30.302 Yes, the presentation aspect. 00:42:30.302 --> 00:42:46.699 So in the wording and also in the tone, I'm

trying to like encourage some healthy skepticism on your part, and I'm being a little bit humble about what I know, even if I'm not saying, you know what, I don't actually know, I'm still giving you an answer, but it's giving you an answer that is also like encouraging, like you might want to look a little further into this. 00:42:46.699 --> 00:42:50.380 It doesn't really work right, you know, like lawyers don't like that. 00:42:50.380 --> 00:42:52.300 That type of thing. 00:42:52.300 --> 00:43:00.378 But I think it's giving a type of personality to these systems, personality in the sense of presentation rather than personas. 00:43:00.378 --> 00:43:11.577 Like we're not trying to pass them off as people or as human, but a certain personality of deference or humbleness that is is missing a lot right now. 00:43:12.378 --> 00:43:18.858 Critical thinking, in your opinion, is that going to go away or is that something that will get sharpened? 00:43:20.137 --> 00:43:26.378 Yeah, I mean, I think a lot of that is up to us as product designers of how we present this stuff. 00:43:26.378 --> 00:43:33.175 You know, our friend Chris Noessel wrote a wonderful book called Designing Assistant Technology that is about exactly this. 00:43:33.175 --> 00:43:41.494 How do we create assistance that amplify judgment and agency instead of replace it? 00:43:41.494 --> 00:43:45.896 You know, and I think we've all had experiences where we have become dumber by using technology. 00:43:45.896 --> 00:43:50.215 I cannot read a map anymore unless it has a little glowing blue dot on it. 00:43:50.215 --> 00:43:56.534 There are definitely things where we're like, oh, I have outsourced some critical thought here and lost some skill before. 00:43:56.534 --> 00:43:58.295 There is de-skilling as a risk. 00:43:58.295 --> 00:44:09.813 So I think one of the real challenges are what are the things that we actually is a skill that we legitimately don't care about or need that is appropriate for technology to do? 00:44:09.813 --> 00:44:20.452 And what are the things that we want to reinforce as we think about introducing intelligent interfaces, as we think about productivity? What are the things we don't want to automate away? 00:44:20.452 --> 00:44:43.010 Either because they are special to us, prized by us, things that we want to, as human beings, continue to do as part of our craft, or things that just require at this point our discernment and judgment, and that we should have people involved in this in a collaborative process. 00:44:43.010 --> 00:44:49.251 So I think that there are a lot of things of like, where do we want to pause and make people think? 00:44:49.251 --> 00:44:53.731 And there are moments of where do we want to put intentional friction into this? 00:44:53.731 --> 00:45:00.849 Just to give a super fast example, you know, kind of of like what Veronika was talking about, about sort of different kinds of presentation. 00:45:00.849 --> 00:45:04.449 We see this in the familiar

things of Google search. 00:45:04.449 --> 00:45:11.088 If you ask, is a lizard a good pet, you'll get different results from if you ask, is a lizard a bad pet. 00:45:11.088 --> 00:45:15.409 You're asking the same information, but the way you've asked biases the algorithm. 00:45:15.409 --> 00:45:18.289 And so Google tries a lot of things around this. 00:45:18.289 --> 00:45:19.329 They will detect that. 00:45:19.329 --> 00:45:21.889 And when you ask, are lizards good pets? 00:45:21.889 --> 00:45:25.409 it will say, also showing results for are lizards bad pets. 00:45:25.409 --> 00:45:32.126 It has recognized that these things are similar but different queries, and it will ask both and give you the results. 00:45:32.126 --> 00:45:34.847 Or people also ask, right? 00:45:34.847 --> 00:45:38.606 Those things of like, you can ask it this way or that way. 00:45:38.606 --> 00:45:42.525 And what it's doing is showing adjacent answers or models. 00:45:42.525 --> 00:45:44.606 There's more than one way to get this. 00:45:44.606 --> 00:45:56.684 And here's a few nearby paths that suggest A, the system is only showing one corner of this area and helping the user to show how to explore more. 00:45:56.684 --> 00:46:03.965 So it's becomes it reveals its bias and it helps the user overcome it and show through it. 00:46:03.965 --> 00:46:05.005 And it's like, oh, that's subtle. 00:46:05.005 --> 00:46:06.684 There's a lot of nuance there. 00:46:06.684 --> 00:46:14.925 How do we think about that of like showing, revealing the shortcomings of the system and helping the user answer those? 00:46:15.724 --> 00:46:17.405 Great. Veronika, what's your take? 00:46:17.804 --> 00:46:22.284 Yeah, I will wrap up by kind of referencing our generational difference again. 00:46:22.284 --> 00:46:29.804 And I've thought about this in terms of literacy, in terms of media literacy and as well as data literacy. 00:46:29.804 --> 00:46:40.681 We've really seen Gen Z grow up in the age of social media and kind of develop this skepticism and cynicism towards social media, towards advertisements. 00:46:40.681 --> 00:46:51.001 I think Gen Z is much more easily able to tell when something is an ad, you know, just because like we're so exposed to it and inundated with it. 00:46:51.001 --> 00:46:54.681 And it's not always a problem, but it does become a problem when it's not disclosed. 00:46:54.681 --> 00:47:03.960 And so I do wonder if Gen Z and Gen Alpha and ensuing generations will develop a similar kind of literacy towards AI. 00:47:05.000 --> 00:47:12.199 And on that note, thank you so much, both of you, for you know joining me this morning and educating me. 00:47:12.199 --> 00:47:14.280 I hope you all had fun as well as I did. 00:47:14.760 --> 00:47:15.400 You really did, Zohra. 00:47:16.199 --> 00:47:16.599 I definitely did. 00:47:16.599 --> 00:47:17.159 Thank you so much. 00:47:17.159 --> 00:47:19.320 And we have a discount code for your listeners. 00:47:20.119 --> 00:47:20.360 Yes, please. 00:47:20.360 --> 00:47:22.440 And I will be adding that in the show notes. 00:47:22.440 --> 00:47:24.518 So thank you for reminding me about it. 00:47:24.518 --> 00:47:26.358 Thank you and have a lovely day. 00:47:26.358 --> 00:47:28.677 Thank you for listening to Inside Tech Comm. 00:47:28.677 --> 00:47:33.798 If there's a guest you would love to hear on the show, please let me know. 00:47:33.798 --> 00:47:38.996 And don't forget to follow and catch every episode on your favorite podcast app. 00:47:38.996 --> 00:47:40.358 See you next time.