← Insights Unlocked
Insights Unlocked
Using AI to streamline and elevate UX research with Google's Alita Kendrick
00:37:29
Google’s Alita Kendrick shares how generative AI is transforming UX research—cutting analysis time, boosting insights, and enhancing storytelling.
Loading summary
Transcript47 lines
- [00:02]
Nathan Isaacs
Welcome back to the Insights Unlocked podcast. In this episode, we're joined by Alita Kendrick, a UX researcher at Google, to talk about how generative AI is changing the game for research teams. From cutting analysis time in half to helping researchers tell better stories and democratize Insights. Alita shares practical tips and lessons learned through from the front lines. Enjoy the show. Welcome to Insights Unlocked, an original podcast from User Testing where we bring you candid conversations and stories with the thinkers, doers and builders behind some of the most successful digital products and experiences in the world, from concepts to execution. Welcome to the Insights Unlocked podcast. I'm Nathan Isaacs, Principal Content marketing manager at UserTesting, and joining us today as host is Michael Dominic, User Testing's Head of AI. Welcome, Michael.
- [01:00]
Michael Dominic
Hi everyone. As always, it's great to be here.
- [01:04]
Nathan Isaacs
And our guest today is Alida Kendrick. Alita is a UX researcher at Google where she leads accessibility research for the Google Cloud platform and explores new market opportunities. With a background at Nielsen Norman Group, she brings deep expertise in design systems, research methods and turning insights into strategy. Welcome to the show, Alida. Lovely to be here. Thanks y' all.
- [01:30]
Michael Dominic
So, Alida, yeah, it's great to have you here. I think maybe a good place for us to start our conversation is to kind of, you know, pick up potentially where we left off in your presentation. So you had come to our Human Insight Summit last fall. That presentation really resonated with our attendees and with our audience. So in it, you shared how AI tools are significantly reducing the time it takes for you to produce insights for your work. Can you help our listeners understand how are you and your team? How are you using tools like Gemini to improve your research process and potentially any updates on use cases and things that you've been doing since that talk in the fall?
- [02:11]
Nathan Isaacs
Absolutely. So I'll start a little bit high level and then we can kind of drill into each one of the phases. And if I'm thinking about like a traditional research process, right, starting with defining the problem space, clearly articulating the problems we're looking to solve and the questions we're looking to answer, really where I'm using generative AI specifically is helping me speed up the creation of some of my planning documents, managing some of my participant communications and stakeholder communications, as well as super lightweight things like developing a project timeline where maybe I have a bunch of times in my brain, but like the use of AI can just help me speed up the actual documentation of some of those things in the more planning cycle related to the specific method Selection and things like that. Again, drafting some of that participant communications through the use of generative AI can speed some of that up and also help me determine kind of some of those study guides or additional resources used throughout my testing. Usually I have like a full script with my participants that I create. That's kind of a common methodology, but this just helps me in getting that time to delivery a little bit shorter in the context of conducting the actual study. Of course, this varies from study to study, but one of my recent use cases with user testing is a great example of how a lot of the AI synthesis that you all use can really help speed up interpretation, which gives me as a practitioner a little bit more time to focus on the storytelling, the democratization of those insights and really pushing for action. And then finally, of course, that like large phase of analysis where I commented of how user testing can play a role. And then of course, when I'm doing methods that may be more interview focused or concept testing in a moderated situation, of course, you know, generative AI like Gemini can really help with qualitative data analysis. I love using this as sort of the halfway point for a lot of my studies. I'll do a check in with generative AI to say, okay, based on all of these transcripts and my core research questions, how am I looking, am I answering those questions or what should I be prioritizing in the second half? So there's so, so many applications from those early stages of study definition all the way into analysis and presentation.
- [04:34]
Michael Dominic
Yeah, that's great. I mean, you've talked about a lot. I feel like so much to unpack there. I think one of the things that, you know, our listeners, they're in similar roles. A lot of our listeners are in similar roles to, to yours and they're thinking about how do we integrate AI into our process. But a lot of times I think before they do that, they're kind of curious about like how much time is this actually going to save me? Is that something that you're tracking and kind of reporting throughout the organization?
- [05:02]
Nathan Isaacs
Yeah, from a qualitative perspective, absolutely. So I'll speak specifically to, to some of my experiences because I've become quite process oriented over the past couple of years and that's enabled me look more detailed at how much time I was spending at certain phases. So even just looking back at my calendar from, you know, three years ago, engaging. Okay. I would have research blocks on my calendar, like two weeks set aside for analysis. Those types of things can actually be sped up into just be A couple of days to highlight those top takeaways, and then really within a week, I can turn around findings. And that's been quite consistent for me throughout, you know, the presence of generative AI and particularly over these last two years, I would say on average, for a lot of my own analysis, particularly of qualitative data, that's cut in half for me. And of course, you know, there's. Whenever we're introducing new processes and new tools into our workflows, there's always that bit of growing pains, right, where you're learning, you're setting up yourself for success, you're figuring out what's working, what's not. And I know for a lot of people in this era of generative AI, that's kind of the hardest part, right, Is just figuring out where, where to start. But maintaining that continuous perspective of experimentation and reflection in the same way as we do with our products will serve you well in kind of this advent of generative AI for our processes.
- [06:26]
Michael Dominic
Yeah, that's, that's awesome. I think you talked about in the analysis phase, you would block off two weeks. Right now you're doing that in a couple of days. Like, that's, that's a pretty meaningful change. Um, what has gaining back that time allowed you and your team to do.
- [06:44]
Nathan Isaacs
More of, Definitely, as an individual, more time spent democratizing access to those insights and sharing them out. It felt like, you know, for a while we were just continuously back to back to study, to study, and it became hard to really allow people to sit on those findings and practice the storytelling. And though generative AI is really good for the analysis, it's not going to go out and tell your story for you. It's not going to sell these ideas, ideas to your stakeholders. So I found so much value in that, in addition to reflecting more on the process side. So, as I said, over the past couple of years, it's really freed me up to think more about the research operations of the programs that I'm owning, which, again, can't really. Generative AI can help. It's augmenting some of that, but it's not leading the charge as it can with some of the data analysis, partly the ops, the relationship building, the storytelling. Absolutely.
- [07:36]
Michael Dominic
Yeah, that's, that's a really interesting way of looking at it. So it's integrating AI into the process of conducting the research, pulling out the insights. But then on the other side, how do we use generative AI to help us share those insights and democratize them out to a larger cross section of our Organization. So I heard you talk about Gemini as being like an important general purpose tool, Like a general purpose AI tool in that process. You mentioned the AI features in user testing as a point solution that you're using. What other tools might be in the mix here that are helping you and your team?
- [08:12]
Nathan Isaacs
Yeah, for all of the program areas that I own, there's a Notebook LM for each one of those. And that is.
- [08:18]
Michael Dominic
I love that product by the way. I use it every day.
- [08:21]
Nathan Isaacs
I love that. It's so cool to hear like two different levels of usage, then of like consolidation of lots of insights and then again that kind of ideation, mind mapping, connecting the dots. I love to hear that.
- [08:34]
Michael Dominic
Yeah, great. So NotebookLM, we've got Gemini as a general purpose tool, again, user testing as a point solution. Anything else that's a common part of the practice. Like any other AI tools that are kind of in the mix there, I'll.
- [08:48]
Nathan Isaacs
Leave it at that for the Google side. But on, as I mentioned, kind of being an instructor at Nielsen Norman Group, I recently developed a class called Efficient UX Doing More With Less. And that really has pushed me to learn a lot of the different AI tools out there. I mean, I had always been quite experimental and actively engaged with what's happening with GPT, what's happening with Claude and different generative models, as well as things like Perplexity Mid Journey and all of that. But through this class I really focused a lot on learning those areas. And of course you get the sense of the competitive analysis, what works best in different contexts. But I think for me, as someone who is not only a heavily active user of these, but also a designer of them, it is really nice to compare those. So depending on the types of tasks, particularly in the case of some of my consulting or my work with Nielsen Norman Group, I'll put it in a bunch. Like I'll put the same prompts in Claude and Gemini and GPT. As a researcher, we're curious, right? One of my favorite things about Gemini, and this is so core to my workflow, is the, the voice experience of being able to just talk to Gemini and have it listen. And I use that all the time for mind mapping. That's completely revolutionized the way that I do my mind mapping and brainstorming because since the beginning of my career I've been that person, you know, with a big Post IT board. I have my Post IT notes and my Sharpie for everything. And I'm kind of going about. But sometimes, I know it's funny to say this but sometimes the physical labor of typing or of writing is just like a blocker to me and wanting to do it like I just want to be able to speak and express my thoughts verbally. And generative AI has just been fantastic for that. And to your point of kind of the writing side as well, getting it used to your way of thinking can really help notice those thought traps over time or help notice where there may be some bias. Because I know as you know, as an industry, we'll talk a lot about the bias that is implicit in these models. But what I don't think people talk enough about is how it can really help you identify your own bias and kind of help correct that in your own thinking, particularly in reporting when it comes to data and things like that.
- [11:04]
Michael Dominic
Yeah, yeah, that's really interesting. So coming back to AI, in the process of conducting UX research, are there any common mistakes that you see UX researchers typically making when first trying to leverage AI into their work and you know, if there's any suggestions you have about how to avoid those mistakes?
- [11:24]
Nathan Isaacs
Yeah, I notice a lot of people just jump in directly and kind of ask their question. And it for sure is a best practice to really define what you're looking to get out of the models. You know, sometimes it's good from the beginning with no context, but to just make it much more likely that you're going to get a quality output that's actually answering your question. The more context you can provide about the AI's role, its objective, the context of your problem space, you can even get as, as defined a specific to find the desired outcome or output when you're in a defined context. Of course, if you're doing more open ended activities like brainstorming potential use cases for a feature or potential research questions in a topic space like information architecture overhauls, things like that, maybe not as much specifying the output, but still defining the role. And that context is really helpful. I'll find some people when they're, you know, really trying to identify areas to use generative AI, they may get into specific of a context at their very first attempt and then receive outputs that don't meet their needs and that kind of scars them away from it for a while. And the whole point with this is really being experimental and I think that, you know, AI from a social aspect has really come about and in some ways scared some people. There's of course the talks of like, oh, it's going to steal my job, it's going to do this to, it's going going to do that. But I think now we're really at that phase where it is, you know, you, you do need to be experimenting. And even if in the beginning it's not giving you what you want, don't bring a fixed mindset to that. Take an experimental lens to say, okay, how can I improve these prompts? And Michael, I know you and I have talked about this, but the use of Mid Journey has been that for me. And so with Mid Journey getting started, I was like, I don't know how to write a really good visual prompt. Like, I'm coming back with things that aren't really fitting my needs. So then I'm like, I'm just going to ask GPT or I'm going to ask Gemini to write the prompt for me, and instantly the quality shoots up.
- [13:26]
Michael Dominic
So use AI. To use AI, Absolutely.
- [13:32]
Nathan Isaacs
Yes.
- [13:32]
Michael Dominic
Yeah. So speaking of the outputs, I think I remember in your talk in the fall, you had mentioned that you approach those outputs with a healthy level of skepticism. Right. So how do you balance the usefulness of the output against making sure that you have that healthy skepticism?
- [13:52]
Nathan Isaacs
Yeah, you have to be so close to the data. Like, for anything that I'm putting into GPT, to Gemini, to any generative AI model, I am the subject matter expert of that research study of the project context. And so, of course, being that person who facilitated all the sessions and things like that, I'm going to be able to know if something's a little fishy, if it feels like a stretch or they're calling out or overemphasizing themes that I'm like, really? That only came up once. So your closeness to the context that you're asking about is so important. It would be like me using a generative AI to be a lawyer for me, and I'm not a lawyer. So, like, it sounds right, it sounds good, but I don't, I don't. I don't have that closeness with that profession to know what's actually quality and what's not. And that's definitely a mistake that people will make early on.
- [14:46]
Michael Dominic
Yeah. Do you think, are there areas of UX research that generative AI maybe doesn't fit in?
- [14:55]
Nathan Isaacs
I think from an experimental lens, always. I think that there's always an opportunity to engage with it, but the usefulness of it kind of depends on, again, how close you are to the data, your overall level of. You want it to sound like you and still fit like you. That's another thing that I'll see from people is you'll be able to tell like this is entirely done with a generative. Like it doesn't sound like you, it doesn't fit, you know, the vibe of the way that you speak or the way that you think. And I think those are contexts where if anything, you might start to see an over reliance on that. And when it's purely generative AI, you know, that's an easy way to, okay, maybe we don't need that job. But if you're layering in your own information, your product context, your organizational context and things like, like that it maintains its value.
- [15:45]
Michael Dominic
Yeah. So I think, you know, we were saying that it sounds like generative A, like a lot of these outputs that people end up taking. Like they just take it. And it's very anodyne what it comes back with. Right. It doesn't have a ton of your personality in it. Are there any, is there anything that you're doing to make sure that it sounds like you and takes on more of your personality when it's giving you outputs, especially when you might end up using those outputs somewhere else?
- [16:11]
Nathan Isaacs
Yeah, I'll give two examples. I'll give a Google example and then an NAG example in the context of Google. I do still do a lot of that speech to text and I'll make it quite particular to say keep my tone of voice, keep my way of speaking because my way of speaking does translate well to my documentation. And I've also been in situations where I have the first draft done and I'll put that in and make it clear to not change the tone of voice or the writing style too much. Just maintain action oriented language, clear conciseness.
- [16:44]
Michael Dominic
What I do typically is I'll give it writing samples of stuff that I've written and say here's me, like here's my voice, here's my tone. This is what I typically sound like when I write. You know, mimic that. Right. I think the problem that I have is my writing is pretty anodyne. So like it doesn't have a lot of character or personality to it. So then ChatGPT just sounds like me and ChatGPT, which is not as ideal but.
- [17:08]
Nathan Isaacs
But I think then that standard way of speaking is the style then. And that's such an approachable, like inclusive way of speaking then as well. And I do exactly what you're talking about for my N and G content. So I recently published an article on fixed and growth mindsets in ux and that was a case where I've published so much for them that my writing style can be quite defined. And so in using Anthropic, I used to claude for that to help me in copy editing that work, it matched my style. And so it is, you know, when people are getting into those situations of more publishing and things like that, even if it's just, you know, your consistent research reports, that's something that you can all provide that as context to the model and say, okay, now, based on what we're working on, improve this content, but ensure it's matching my way of speaking and my style. Because I think you will be. I always think about, especially now in being in a large corporation of just your brand as an individual individual and as a contributor to your team. And your brand is often seen through the deliverables that you're sharing. So when someone goes through kind of such a stark revamp around this time of Generative AI, you're kind of like, oh, okay, like how much reliance do you have on this? And I think we might be getting to that point where we're like, you know, we want to make sure it's a merged view, it's the AI being helpful, but it's not leading the creation of all of this work.
- [18:31]
Michael Dominic
Yeah, 100%. You touched on something earlier I'd like to come back to and I think this was something that you brought up in your presentation in the fall and that that is really around the topic of maintaining objectivity in your UX research. I know it's a goal to remove your bias when you're pulling out insights and when you're trying to really report on what your users are telling you. How was AI helping with that process?
- [19:00]
Nathan Isaacs
Yeah, I love that because AI has really changed the way that I take notes during my sessions before Generative AI. And I've had the same research note taking template since the start of my career. And it was quite simple in that it was a timestamp, a question or a theme that I'm noting, and then the actual note column and then a separate entire column for tags of like question marks, for follow ups, exclamation points and things like that. And when I was taking notes back then, it was like everything. It would be quotes, it would be a key takeaway, it would be they clicked this and this was their workflow. So it was so many different levels of note taking because we want to get it all in now with the use of Generative AI, I'm already getting transcripts so I can really focus on whatever quotes I'm documenting. They're probably going to make it into the report or that's like a really nice high value quote. If I'm highlighting an action, it's probably a specific theme that I'm like, okay, this is going to be a top takeaway. So during the session I can be much more focused because I'm not scrambling to take as many notes because they're already being tracked. And I think from there we take those and then I have basically a little, a medium sized post to know where I have all the top takeaways from that. And so as I'm analyzing that data through the use of something like Gemini, I'm saying, here's my session by session topic takeaways, here's all the participant transcripts. What are you seeing as the main takeaways from this? Because again, you know, everyone's biased. Sometimes people will talk about bias as like, oh, it's this bad, this bad thing, right? Ooh, bias. Really? It's just our perspective, right. Which is informed by our past experiences and things like that culture, all that good stuff. So in the context of our findings, I may say, okay, I have an image implicit bias that like, I think this is going on. And so I make note of that in my own interpretations. But really when the model goes through it, maybe it's like, yeah, that's a little bit there, but it's maybe not as there as prominently as you're connecting those dots. And so I think that's such a solid way where we can start to say, what are we seeing? What is the model seeing? And kind of merge that story together. And again, sometimes you might be in these contexts where, yes, the model knows about the sessions, you know about the sessions, but you, as the leader of this work also understand your business context. So you might be in a situation where, sure, the generative AI is suggesting X, Y, Z, but you know, from a business perspective and your organizational perspective that that's actually not connecting in the way that it should. So this is where again, that storytelling is so important and you can't just take it as face value of, okay, the model said this, so this is our source of truth. It still only has limited context compared to everything that you know.
- [21:45]
Michael Dominic
Yeah. Does that connect back to the point you were making earlier about democratizing these insights? Yeah, I think that's a really, it's, I mean, democratizing research, Right? Like that's such a big thing in UX research and it has been for, you know, the last several years. It's interesting because I talk to a lot of our customers about this as well. It's like, okay, you've got all these really robust democratization programs going where you have non researchers getting involved in the research process or in the synthesis process, or all of it, right? And there is an interesting role, I think, that AI does play in that process. Do you have any, like, more like specific examples of how your team is doing that? How does AI fold into just the larger democratization efforts in your research process.
- [22:33]
Nathan Isaacs
Early on, to your point of kind of seeing multiple perspectives too, even in the study planning, I can say like, okay, here's my research plan. You're a uxr, but like from a product perspective, what questions might you be wondering? From an interaction design perspective, what might you be wondering? And even I think that level of foresight and proactivity, when your stakeholders then go and review that doc, they're like, oh, okay, this is well rounded, it's well structured in a way that it may not have been had that not been top of mind. And you know, without the use of generative AI, those types of tasks can say, oh, it's going to take me so much time. I need to really sit down and empathize with my PMs or with my interaction designer and this and that. But that can definitely speed up against some of that proactive side of things later on. One of the things that I've noticed, particularly at Google, is there's so much research being done, like incredible quantities of research, and things are moving so fast that sometimes that research just gets buried by, you know, what's hot and new in the same way that rendered of AI buried a lot of that. And so one thing that I'm using generative AI for a little bit more right now is more creative views of, you know, taking reports that I had and repackaging them in a more interesting way. At the user testing conference, I know there was a talk about this where one of the practitioners used tarot cards to help with the storytelling. And I thought that, you know, that example has stuck with me so much because it's, it's such a beautiful way to again repackage our standardized research reports in a way that's more interesting because in this era of just, you know, big data is beyond big data now and we're stacking generative AI on top of it, and things are kind of starting to look the same. We want to make sure that we are being those creative storytellers and really insightful storytellers and generative AI can help with that as well.
- [24:25]
Michael Dominic
The democratization thing I think is really interesting too, because from, you know, I've been pretty close to a lot of democratization programs being here at user testing over the last several years. What I see is that it is a, a lot of these programs lean heavily on UX researchers, like expert UX researchers. Right now what I'm seeing is those researchers are starting to create custom GPTs or gems, right. That, that are, you know, that reference point for conducting like really good research. You still want to keep that process on the rails, but now there are new possibilities to maybe outsource a little bit of that expertise to, to AI and free up UX researchers to do the work that they're really good at.
- [25:11]
Nathan Isaacs
Absolutely. Yeah. I see so frequently that researchers backlogs are just massive and there's always more studies to be done than researchers to do it. And so particularly through, you know, agents and more autonomous and just artificial intelligence, we will be able to do more research. But it's not just the fact of doing more studies, it's also increasing the impact from those studies as well.
- [25:36]
Michael Dominic
Yeah, 100%. Yeah. I think that like you can continue to bring AI into the process to help create those efficiencies, but I don't know if that backlog ever gets smaller. Right. Because the more time that you have to do that really interesting in depth research, the more you can do. Right. So yeah, I don't know that that's going to go anywhere anytime soon. And I think that is the possibility of AI and UX research.
- [26:06]
Nathan Isaacs
Yeah, 100%. It's so interesting to see practitioners experimenting with things. One of the things that I heard most commonly from the user testing conference where people coming up to me and just saying it's been so hard to get started. And this was back in October. So you know, people then are still struggling to just get started and people really commenting on how helpful it is just to have that prompt cheat sheet to say, okay, what are some specific prompts that I can just get started with and not have to think about it? So I think organizationally if you are an organization looking to really up level the use of generative AI, it's definitely top down. You know, Google was great at this. They ensured, you know, people are doing their AI trainings and things like that and I think that level of, of top down really gets people on board to say, okay, this is something that I need to prioritize now. My leadership is very involved and then just keep up that momentum of experimenting. I even have a doc for myself. I have lots of scratch pad docs for my different programs and I have a doc of just that is like my go to prompts. One that I know I've talked about with you before is I adapt so many things from consulting just coming from, from that arena. And one of the things that I absolutely love for my surveys is asking the generative AI to ensure the options are mece mutually exclusive and collectively exhaustive. And that has made so many of my survey questions much better in a way that like would have taken me hours and additional research and things like that where I'm like, no, this makes sense. Like this is good, solid. Okay.
- [27:46]
Michael Dominic
Yeah. So Alida, I have to be honest with you, it doesn't surprise me that there's smart people at Google that are really embracing all of this, like Google being the creator of one of the frontier models. So I would put, I would say Google's a little bit of an outlier compared to most other companies. But let's take a step back here and let's imagine, you know, I'm a UX researcher at a company that maybe has not been embracing this change or really has not been leaning into that change. We don't have access to enterprise level models. Maybe we're being told to, you know, just use our own personal accounts to, to Gemini or ChatGPT. What is a good starting point for that person? Like that person doesn't have prompt libraries that are being created for them. What do you think a good starting point for that person would be?
- [28:32]
Nathan Isaacs
The first thing I'm thinking about are what are those tasks that you hate doing? What are the tasks that are taking you a long time? And that's the perfect incentive to experiment with generative AI. I'll be honest to say one of the things that I don't like doing and spending too much time on are my emails and like preparing, wrangling my stakeholders to get their approval on, on a research plan or something like that, and even just drafting some of those communications, whether it's a chat message or an email. It's just something that honestly I don't like doing. I like to focus on my research. I like to communicate in person, all those things. And that was one of the first areas that I really started to being like, no, I don't like this. Like I'm just going to use generative AI, get me a baseline and then I'll adapt it to my voice. And because it's a task that I don't like doing, generative AI will speed up to 80% of the time for me to complete that task. And so really when you're getting started, think about those things, reflect on your own processes, what you don't like doing at work that could be improved or optimized through the use of an artificial intelligence. And start there.
- [29:40]
Michael Dominic
Yeah. So what comes next then? So, you know, I guess like another scenario would be like, I think what we're just kind of getting at here is a crawl walk around approach. Right. So like the crawl version is finding those low level tasks that take you time, you still need to do them. AI comes in and helps you improve the efficiency of that. Those are all super valuable use cases for AI. I think the most valuable use cases for AI is using it as a thought partner. And you've kind of talked a lot about different examples for that. So how does someone go then from I'm using it to write emails, I'm using it to create blog posts, or I'm using it to write screener questions for me for research studies. How do you go from that to these really high value unlock use cases?
- [30:30]
Nathan Isaacs
Yeah, it is such a habit to build. And truly I'll be like, I'll find myself spinning on something in my head and then think, oh wait, I should just let me ask Gemini, like what? What's Gemini going to tell me? And so I think it really is setting yourself up with that habit to say, let me just go and see what it's going to give me. It's not always going to be something that I'm going to end up using. But even just getting to that habit of how can this help will really set you up for success in the long run. I think if we're thinking, you know, that single point of interaction level with it where yeah, we may be using it to help us with communications or with writing or ideation of something, the next kind of area then is reflecting at more of a project or product level. What are those opportunities where generative AI can come in and help me? And from there you naturally do shift into this relationship with generative AI where it's just, you know, those are the people that are using it multiple times a day. It's second nature to just go and ask. I've got Gemini, GPT and Claude all on my phone in addition to a bunch of other AI apps because it is just part of my relationship with technology now. But you don't get to that overnight and you don't get to that just by using them in isolation. You really start to feel that connection and the power when you follow it through, you know, into a project who from defining your research into the presentation of that research. Like going through the full cycle is when you'll really start to have those light bulb moments of oh, I see what worked there. And this is where truly being such a reflective individual helps with that. Just like any sort of project retro, postmortem type of thing, adapt that same thing to your usage of AI. Do something like a quarterly check in with yourself to say am I doing more? Or what worked, what didn't work? And you can do that from your organization level as well. Just to say we want to see this many queries that people are using, but it really is kind of that habit, that experimentation and that reflective nature to say what's working, what isn't.
- [32:34]
Michael Dominic
Yeah. So AI is proliferating like crazy, right? The models are constantly getting better. People are using them more and more for really interesting use cases which you've talked a lot about. What excites you most about AI, whether it's in UX research or just in general, what are you, what are you most excited about?
- [32:55]
Nathan Isaacs
In general, I'm just so pragmatically optimistic about it. Like I, I'm the type where it is really balancing the hopefulness with like the realism of it. So I use it in so many different contexts from like personal things. I've been redecorating an apartment and saying like okay, I want jewel tones. How can I match this rug with photo uploads and things like, like that. With some of my work on the side, you know, crafting examples for my class on efficient ux, building out thought experiments for that has been just fantastic. In addition to just my 9 to 5 at Google and really embedding generative AI all throughout that I would say I'm optimistic across the board. One of the things from an industry level and I've always felt this way about generative AI since, you know, the past years, couple of couple of years when it's become quite present in our day to day communications. My grandmother knows about generative AI now, but I've always kind of felt like the chat experience would be short lived. That was never something that I was incredibly passionate about of saying okay, chat's fine, whatever, but like for a standard consumer base, I just want a smart interface. I want more proactive recommendations and more contextually aware knowledge of, you know, all those things that we talked about like your business context, your personal context, your relationship with your stakeholders, almost that like second brain view. That's what I'm really excited about. Just you know, over time with generative AI from an industry perspective is we're truly only scratching the surface. And right now we're seeing so many of the same types of use cases. But as people get more experimental, I think we'll progress more into some creative problems to be solved through the use of artificial intelligence.
- [34:44]
Michael Dominic
Yeah, that's interesting. So if chat is not the interface in the future, what would be the interface like? What are, what is it? How are we interacting and engaging with those tools?
- [34:55]
Nathan Isaacs
I think chat can always play a role in the same way that search has kind of always played a role in experiences. But think about more personalized views of just where we have so much contextual understanding of, of what you need. How can we serve that up to you proactively? In the same way of, you know, Amazon does this with their ads. There's a ton of AI built into the way that Amazon.com is set up. Do you always need to search for what you need? Sometimes it's right on that homepage because the data is so good and they know you so good. Whether it's something like, you know, maybe it hears that you're redecorating your apartment. So now you all are going to get up on of ads for things like couches or it's your actual behaviors. That contextual understanding can help our users accomplish their goals more quickly rather than just requiring a bi directional chat interface.
- [35:52]
Michael Dominic
This has been a fantastic conversation. So for show notes, including links to what we've talked about, visit usertesting.com podcast. Alida, thank you so much for being on the show. This was a really, really awesome conversation. I know our listeners are going to love, love hearing this. So how does someone learn more about you? How do they learn more about your thought leadership? You know, even the work that you're doing at the Nielsen Norman Group, where you know, where can they find out more?
- [36:20]
Nathan Isaacs
Yeah, so great to be here. You guys can check out my NNg profile for all my publications and the classes that I teach. The efficient UX1 I think is particularly so related to a lot of our conversations today about just reflecting on our work that identifying some areas of optimization for us and experimenting with artificial intelligence. Besides that, of course, I'm available on LinkedIn where I post random things like the cheat sheet that I developed, which we'll also post a link to in the show notes.
- [36:50]
Michael Dominic
Fantastic. Thank you again, Alida.
- [36:52]
Nathan Isaacs
Thank you, Michael. Want to keep the conversation going? You can find the show notes at YouTube if you haven't already. Don't forget to follow us on Apple podcasts Spotify, Overcast or Google Play, so you never miss an episode. And if you enjoyed today's show, please share it with a friend or leave us a rating and review on Apple Podcasts. And until next time, this is Insights Unlocked, an original podcast from User Testing.