Anthropology's Key to Robust AI with Professor Martin Holbraad

Welcome to episode 3 of the Mahesh the Geek Podcast. Today, I'm joined by Professor Martin Holbrad, a leading anthropologist at University College London, the Director of the Ethnographic Insights Lab. He's one of the central figures in what's known as the ontological turn in anthropology, a movement that doesn't just study human cultures but retains the very concepts we use to understand them. For those less familiar, anthropology is the study of how people live, relate, and make meaning, always in context. It's a discipline grounded in ethnography, the deep, often immersive study of people in their everyday environments. And that kind of deep attention is exactly what we need in our current moment. In an age where technological advances like AI are colliding with an increasingly complex social, cultural, and political landscape. At the Ethnographic Insights Lab, Martin and his team of ethnographers work at the intersection of academia and industry, bringing anthropological thinking to bear on the toughest human-centered challenges in both the public and private sector. That's why we partnered with Martin and the lab on a recent project exploring AI and police incident reporting. That collaboration forms the heart of this episode. And along the way, we get into some big foundational questions about anthropology, technological systems, the nature of human knowledge and memory, and what it means to design for real people in the age of AI. The discussion turned out to be so rich and far-reaching that we have split it into two parts. In this first episode, we lay the groundwork, exploring Martin's approach to anthropology, take on conceptual innovation, and what it means to reimagine design through the lens of ethnography. So let's get right into it. Can I start off by just asking a little bit about your background? How did you get into anthropology and what got you interested? I actually had a very kind of a moment of epiphany. I was a philosophy major as an undergrad at the London School of Economics, which is a famously kind of dry philosophy of science place. And I really enjoyed my course, but I found it a bit empty of people, the kind of stuff that philosophers in Europe talk about here in Britain and the United States, the analytical tradition that don't really go into these things, they kind of turn philosophy into a kind of logical gain. So I had a bit of a brain wave at some point in the middle of the night as a third year undergraduate where I realized that, hey, hang on. Philosophical problems actually feature in a people's everyday experience. So, for example, since the ancient Greeks, there's this distinction between the one and the many. Is everything one, as Parmenides said, or is everything multiple as Heraclitus said, right? And you can think of Plato and Aristotle and the whole development of Western philosophy as a rumination on this kind of dual possibility of the one and the many. And there's lots of really interesting stuff written on this in a very abstract way. And I was thinking, think of a roulette. So people gambling, literally putting money, stakes, sometimes their houses, their livelihoods and so on. On a technology that is split in a digital way into 36, 37, 38 slots. And then the continuous one-like motion of a ball kind of finally selects one of those things. So the many options and the one movement is basically this kind of philosophical contrast between continuity and discontinuity, the kind of stuff that philosophers go on about. Acted out in real life, people staking money on that, right? So couldn't one approach philosophical problems by engaging with real life, right? Couldn't one say something interesting about the one and the many by spending time with gamblers? Now, I didn't know it at the time, but that's exactly what anthropologists do. They spend long quality time with people just working out what makes them tick. What are the assumptions they're making in a situation? What are their practices? How does it rip out a different bits of their life connect to each other and so on? It really kind of just being with people, right? So my objective was to do that in order to speak back to those big philosophical questions that I was working on as an undergraduate. So I think my whole trajectory ever since has been that kind of tag between philosophical conceptual questions and real life tangible concrete experiences. Which I think for me is a definition of anthropology, that's what we do, right? So maybe to make it concrete, how would you define anthropology for a lay person? So this is an old chestnut, people have tried to do it in different ways, right? So I would say anthropology is concerned with the social and cultural dimension of the human experience. But that we share with many people. Methodologically, we do have something distinctive in the way that we look at social and cultural experience, which is the word ethnography, which has the prime methodology that anthropologists use, which famous American anthropologist called Clifford Kietz famously called Deep Hanging Out, which I think is a great definition, because you literally hang out with people for a long, protracted periods of time to try and go deep into the underlying motivations of their actions. And the structures and the logics that underpin them. So what I distilled over the years as the kind of signature of anthropology as an approach and as a discipline, is the fact that yes, it looks at social and cultural aspects of the human experience. But in so doing, it always involves the people who are involved in those phenomena. So if I want to understand how gamblers, as I was talking about before, how they go about thinking about their activities as gamblers, I spend time with gamblers and I ask questions like, what is gambling to the people involved, right? What is the significance of the roulette from their point of view? What does time, the experience of time, look like from a gambler's point of view, right? And then by adopting that point of view of the person involved perspective, I then recursively or reciprocally, if you like, define the coordinates of the way that I study the problem. So there's a kind of recursive or reciprocal relationship between the person doing the analysis and the thing being analyzed, right? And that reflexive interplay between how I think of a problem and how the people involved in that problem think about that problem is that special turn of thinking that anthropology involves, right? Interesting. Interesting. And as you sort of project this to perhaps not just an individual gambler, but how that gambler perhaps interacts with society, how do you think anthropology attempts to think through the culture of a society and the way they operate and the way they think versus the individual itself? We would always say that those two things are functions of each other. So what a gambler does in a casino, let's say to use that example, is a function of social and cultural forces, assumptions, schemes, logics that are manifest in that casino in his own behavior in their own behavior, right? So what the gambler takes the role to be, what the moral evaluation, for example, of being a gambler is how they define themselves vis-à-vis that, you know, perilous kind of image of the gamblers being, you know, a kind of volatile and possibly addicted person and so on. All of those issues that would come up in a netographic engagement with gambling are things that are social and cultural. So we as people are not individuals set against society or culture. We are expressions of it and modifiers of it. We are created by society and culture, and by the way, those are also open questions. What is society? What is culture? It's not a settled question. It's a question that you can open up at my graphically, right? But we are conduits or expressions of social and cultural forces and schemes and so on, but we also have reciprocal effects upon them, right? So the idea of creativity, of individuality, of uniqueness, of experience, and so on. So we are attentive to both, if you like, the conventions of social and cultural life as anthropologists and the way that people's behavior, as assumptions, interactions, relationships, and so on, manifest those conventions. But we're also attentive to the capacity of human beings, possibly uniquely, to invent those things and reinvent them and transform them. Interesting. So as we think about maybe mapping this to this notion of public safety organizations and we had an opportunity to collaborate with you on a specific research effort there. How did you think about a public safety organization before you embarked on this effort and what changed as you dug in and understood the individuals and the structure of the organization? I mean, may I say this, this project has been so much fun. I said it many times and I know that may sound like trivializing. I'm a bit scared of using the word fun because people here should be serious work. It shouldn't be fun. It's so much fun. And one of the reasons I think that it's been so fun is the optimal combination for me personally, of tapping into a world of practices and phenomena that we've all grown up. With all watched cop shows on TV and you know, emergency services are in every Hollywood blockbuster that you've ever seen and so on. So ever since I can remember myself, I have an image of what a precinct in downtown New York looks like or whatever, right. So you've got that kind of the allure of the mediatic charisma of law enforcement of, you know, emergency services and so on. Without in any way wishing to minimize anything, it's just a very kind of big giling world, right. It's a kind of cinematic almost world car chases, you know, you have all of these associations. So finally, I have an opportunity to kind of engage with people who actually do this as a bread and butter. So that on one side and on the other side, and this is where the anthropology comes in, the flip of some of those assumptions. So flipping some of the kind of standard assumptions, which if I may say so in the interviews that we conducted with law enforcement officers in, you know, as part of this project, the police officers themselves sometimes shared in, right. So they themselves are kind of buying into this discourse about what this is, right. And then looking at the actual practices and structures and so on that we were talking about and finding the other things are going on. So to give you the example, and this is, I think, one of the headlines of the study that we did, right. You go in thinking, okay, the glamorous bit of police work is the car chase, it's the solving the crime, it's the putting the bad guy in prison, you know, going into, you know, being the very physicality of turning up into a high pressure situation and being able through your training and sheer talents to kind of deal with it. Right. Of course, that's all part of it. The assumption then is, and we hear it all the time in the media, paperwork, report writing, what a drain on resources, what a, if only we could just not spend time on that, you know. And just have more people on the beat, you know, more people out on the street to the constant. This is a constant narrative that not least politicians are kind of responsible for propagating, right. When we spoke to these police officers, sometimes as I say they would say, oh, I'd much rather be in my, you know, patrol car than being in the office writing reports, you know, there would say things like that. But then when you actually went deeper into the conversation, you realized how incredibly important police report writing was to them, right. In the end, what I, the image that I kind of emerged with is that actually reports are the currency through which police officers measured their professional competence. And I should add, have their professional competence measured by others, right. So, you know, one thing that became amply apparent is that most crucial of all is the capacity that you have to write good reports, which are a reflection of good police work that gets you promoted through the ranks. Right. So actually police officers have a massive vested interest, not least self interest when it comes to personal promotion and career trajectory in having well written robust future proved reports, which is what we were trying to help them achieve through these, you know, through this project. So from a position of our reports of boring, you know, talk about, you know, the car chase to a position of appreciating, of course, it's about the car chase, but report, contribute, report writing contributes to the police officers capacity to do their job well and is something that is massively important to them and a source of pride as well and frustration, of course, when it's not done well. So suddenly, the, the meaning significance and daily importance of report writing really became a 3D kind of experience for a certain graphically. And if you look at the tools available to officers today to help author the report, some of it tends to be pretty basic, it is the pen and paper approach. There are data capture tools, various sorts, it could be audio recordings, video recordings, et cetera, witness interviews that help with providing the context to author those reports. Where do you think the line is between the perspective that the officer, the human needs to offer because that drives their action and perhaps drove their set of actions versus just an objective gathering of facts that then just become compiled in a report. Yeah, well, it's very much a combination of the two and it's very interesting how the officers were caught, you know, that we spoke with constantly attacked between these two things and I think it's, you know, it's not an either or right. So, you know, it's actually, you might say part of the skill of being a good police officer is being able to do both the objective and if you like the narrative, the persuasive, the rhetorical, the subjective, right. And not turn that into some kind of zero, some gain, right. So use your, you know, acute training in reading a scene in knowing what questions to ask, recording correctly all the facts and all the relevant items that need to be there, right. And then mustering that into a persuasive narrative, I remember one of the officers that we spoke with said report writing is painting a picture and when the person who reads the report reads the report, they should see the picture and they should know exactly why you did what you did by seeing that picture, right. So that's a really interesting statement because it tells you about the important picture to pick something so it has to be factually accurate, it can't be just like some random, it's a picture of something of what happened. It is never the lesser picture and it needs to be done well, it needs to be a narratively kind of complete and persuasive as I say document, but it's also, it also has a kind of moral quality to it in the sense that the officer, there's an anxiety underlying this statement that the officer might be challenged. That the officer's judgment is on the line and so on and what's at stake in writing a good report is that whoever reads that report should agree with the officer that the line of action taken in this instance was the correct one, right. So as I say, it's performance of professional competence and a communication to the reader that I've done my job well, right. So it's all of those things together, so it's objective, it's subjective and it's moral if you like, right. And that's the kind of richness that ethnographic engagement provides and once you realize that that has very important implications for, for example, how you think of any of a technological solution to these problems, right. When you see the proportions that the issues at hand have for the people involved with them, then your imagination as to how to help those people is triggered in different directions. Now, there's a series of things that the documentation process aims to do. So one is to serve that the virtuous cycle for that individual officer. It is also supposed to serve in some sense as evidence of what happened, perhaps for someone else to facilitate an investigation of the set of events. So there's a level of structure and a level of things that need to go in there. Some that are really process oriented to facilitate somebody else's work and there's a bunch of things that are perhaps really to serve the self improvement processor of that individual itself. So if I drag a cursor to say, okay, this is where this is the minimum amount of perspective or information that an officer needs to offer. And this is the remaining structure that perhaps is mechanical in nature. It could be making sure the spelling of the words are right, the prose seems coherent in some way. Where do you draw the cursor, where there is in fact some assistance that is possible through an AI or through some passive observer, a Goobstrider versus the input that has to be provided by an officer. Yeah, yeah. I think that line needs to be drawn situationally, but I think it's there. There is no doubt that some things are just facts that if you can get them. And often the addresses and times and so on that need to be repeated on four different forms. If you can find a way of automating that, you know, everyone's going to be happy for that and it's not very difficult. But I think it would be a bit of a failure of imagination if we left it there, right? I really think it's the metaphor that's the problem here, right? And it's funny, I should say that as an anthropological metaphor is that full of meaning and rich and so on. So we like these things to think about. It's right to metaphor, that's where the problem is. And I think that's a failure of imagination in this context. And that's what really I think came out of this project when we were thinking about, you know, the what ifs and you know, how might we use all that kind of stuff, right? It's like, well, what if we stop thinking of AI as plugging the whole of the writing problem, because that's not where the problem is. The problem that we came up again and again against is the ping pong between the officer on the front line and the supervisor has to read, you know, 3040 reports and basically make sure to protect. I was going to talk about duty of care, right? Not least to protect the officers under their command from the scrutiny of complaints, procedures of judges of all of the things that you mentioned before, right? Which is so, so important. And the fear of that humiliation is something that came up again and again of being shown not to done your job properly, right? So typically what will happen and I got this complaint, you know, it was funny. I always ended up feeling that I was talking with fellow academics with his kind of 50 year old, you know, police officers sitting in, you know, wherever and some rural part of the United States or some urban part for that matter. Talking to me much like, you know, a colleague down the corridor might lament about, you know, our students these days, they don't know that one example that one of them said, they don't know the difference between a tenant and a tenant. What are you writing about people renting flats? Are you writing about their logic, you know, this kind of thing, right? These people have master's degrees, you know, like, how can they not know, you know, this kind of thing. So, and there's lots of stories about, you know, five iterations, six iterations until they got it right. But it's really important because if they don't get it right, you know, we can't submit this report. What if we thought of that process as the place for efficiency, which is up to one of the drivers of these innovations, right? So, not think of the initial ghost writing process and how we can relieve the officer from having to do writing, but how might we relieve the supervisor from having to ping it down so many times, right? So, what if the AI tool was effectively a supervisor by proxy? What if the job of the AI tool that you developed were a probing of the inner logic and the robustness exactly the kinds of questions that a judge or a supervisor or a promotion panel or indeed a complaints panel, right? My ask about a report. That's what an AI tool can do extremely well, right? Rather than being imagined as a kind of aid memoir effectively, right? Which is surely the imaginative way of using it, I think. Interesting. Interesting. And so then maybe it's one more question on this, this general topic, and it probably is more of a memory question than it is an anthropology question. But the presumption sometimes also is in pointing in perhaps another flaw in the ghost writer model is that the ghost writer is also a flawed creature. It is the observation is not entirely factually documented in whatever the ghost writer writes. And the presumption in this model is that when the officer gets to read what the ghost writer has written, that they're able to easily spot the mistakes, the issues, etc. But I think what you're research indicated was that's actually not true. In fact, it is quite the opposite that happens here. And maybe you can talk a little bit about that. Yes, I mean, I'll talk on this by virtue of working in a kind of interdisciplinary team with a with a psychologist of memory. She was a pretty happy Rothal and also her name and she was fantastic and she taught us all about the dangers of what's called memory contamination, right? So the degree to which memory is susceptible to hallucinating if you like to use that term for a human in this case. So if it if prompted in the wrong direction. God, do you remember was a Pepsi or Coca-Cola that the suspect was was drinking? Oh, I think it was Pepsi says the other person. Yeah, it was Pepsi, wasn't it? Even though it was actually Coca-Cola, but the fact that someone said it, I think it's Pepsi immediately skews you to thinking that this and actually in your mind's eye seeing, you know, a kind of Pepsi. Other brands are available, but dear listeners. But so hallucination or indeed a kind of summarizing not incorrect, but kind of glossing over of detail, summation of that an AI tool might produce is very likely to skew the office's own subjective experience of memory. Right. And that is well documented. And psychology, this is not it's not a controversial thing at all, right? And so there's a there's an interesting model that you're pointing to here, which is the. The ghost writer model is almost what I perhaps crudely refer to it initially as someone who is writing alongside the officer with the sole purpose of helping document what happened from that person's perspective. That's one model. The other model is that person is not someone who's going to ghost write your report for you. But instead, as you're writing your report, start to question, did that really happen? Are you including the right piece of information? So it is someone who observed just like you did, but now comes in as someone who can identify perhaps the cracks of the report or something that is questionable or something that is unclear. In some fashion or perhaps missed in some fashion and prompts you to make those changes. In that in that model, how does the virtuous cycle get reinforced versus it being sort of this division that you refer to previously? Because what you are reproducing here is exactly the process of supervision and learning and training, right? Effectively, what you've got is a powerful, powerful tool for further training, further attuning you, further teaching you how to read a scene. Why didn't you ask that question? Did you notice that? One of the other recommendations that we made was, and again, this was Hattie, the psychologist on the team, her kind of very valuable input, was the use of what's in psychological literature is called a cognitive interview. So another way that a professional psychologist used this as a way to enhance accuracy of memory report in, for example, witness statements and also in criminal proceeding context. For example, one of the four or five different methods involved in cognitive interview, one of them is to recount the incident backwards. So rather than drafting a kind of journalistic narrative that starts with the beginning and then goes forward, you start with the last event and then you say, what happened before that? Now that is an unorthodox way of recounting the story, but in a highly effective way of recuperating memories while they're fresh, right? Or even, you know, a few days later. So an AI tool that effectively emulates a cognitive interview is it really, really boosting the kind of virtuous cycle that we're talking about adding a new dimension to it, right? So it's not just that we can emulate what supervisors are already doing through their vast experience in training the younger colleagues in their care or in their in their charge. But we can also add new dimensions to that training process on the job, right? We're not talking just about officer training before they start, but as part of compiling a report on the job is lifelong learning that is involved in any professional expertise, right? We're all learning all the time, this conversation we're both learning from, right? So we're trying to enhance that virtuous cycle by thinking of ways to use the power of AI to feed into it rather than to track from it. So maybe almost if you take a take that concept and now expand it out a little bit beyond just the authoring of reports, more often than not now, the predominant way in which AI is applied to a variety of tasks is for it to be almost a bit of a personal assistant. It is somebody who is alongside you readily accessible has access to perhaps a knowledge pool that is much higher than what you have in memory, but at the right moments in time you can interact with it, perhaps it can help you document, perhaps it can do multiple things. Wrapped in all that concept is, okay, there is this constant notion of, is there fairness, is there bias, is there trust? The notion of fairness, bias and trust tend to be very universal, or at least the assumption tends to be that it is very universal, but it may not be that case. So how do you think about it as an anthropologist? What is fairness, bias and trust in this context? Yeah, I mean those are amazingly important questions and I should say that, you know, we've got to do another project on that, because that is exactly, you know, if you like in the project that we did, we asked the question, what is the report and what is memory, right? Those are the two kind of driving questions, and those by the way are ontological questions, right? What is the beginning of an ontological question? What you're asking now is about the ontology of fairness, the ontology of trust, the ontology of the social in some way, because the social is built on trust, right? You and I have to trust the editors and agree to be having the social interaction that we're having right now, right? So what is the social, what is trust, what is fairness, and so on, are central kind of ethnographic questions that we would need to unpick as a project? I think what is relevant in this, I mean there's so much, this could be another house conversation, and there is, if you look at the material culture as we call it, of AI, it's always magical in some way, right? It's always some kind of sprinkly little icon in your desktop that, you know, and that is almost like the kind of branding or the, you know, the comms or whatever you want to call it, the marketing of the product, invokes magic. And magic of course is there's, boom, right, this black box, right? The last person you're going to trust is a wizard or a witch, right? You're not going to trust them because you don't know how they operate, you don't know what they're doing, and we all know black box question in, you know, the explainability of generative AI is a big problem that we don't, it's not even reproducible, you ask the same prompt twice and it gives you two different results, right? So we've got, we've got an issue there, right? So I think that's the area in which you've got a friction between the technological kind of potentials, and the requirement for trust, as you say, which is so central, and of course questions of bias and all of that, yeah, I'm into that, right? We need to find ways, I would say, and I think your principal tangibility is a stab in that direction of lifting the veil, right? I'm not saying just to the technology because I can't speak to that, I don't know if it's possible, or if it should be, or even the commercial stuff, right? But in the human technology interaction, in the pairing between the human being who's doing the prompting, and that, as you say, acutely powerful research assistant or assistant that is the AI tool, right? In that relationship, we need to bring things to light, to open them up, not to be veiled, right, in this way. So I think if you start, you talked about anthropomorphizing before, it is powerful, how anthropomorphic it is, right? To start thinking about the relationship between a person, and an LLM, or an AI tool of any kind, as a social relationship, and I wouldn't be surprised if in 10, 15, 20 years or even earlier, that will start having legal implications, will start thinking of these tools as quasi-persons, as we do with corporations already, right? So we already know that we're treating different non-human entities as persons, right? Some would use that as a definition of animism, by the way, to treat non-human entities as person, as the definition of animistic cosmologies and anthropology, right? So this, and the serious animism, if you like, of treating AI tools as quasi-persons, and all of the implications for trust, commitment, lack of bias, challenge, accountability, all of the things that we would expect in a human-to-human interaction, are going to slowly start to be transposed into that relationship. So for me, it's an absolutely fascinating new frontier for humanity in that sense, not just in this engineer's kind of imagination of all the amazing things that we'll be able to do, but all the new social forms that are going to emerge. That's really exciting. I mean, you would expect that from me, because I'm an anthropologist, I like social media. And now I have these AI agents, so to speak, these entities that I'm talking to, and they're part of this network, and there are actions that we are performing as a whole that are influenced now very much by this notion of what these AI agents are able to influence you to do. They come with a certain level of power, perhaps, very different than what it would be if you were interacting with another human in this network. How should we be thinking about this, especially in the context of really mission-critical set of teams that are operating to save lives, protect things and people, etc.? How should you be thinking about designing in this sort of actor network, where the actors are not all humans? Yeah, well, it's the kind of, what's it called, $64,000 question in some ways, and of course, I wouldn't presume to have the answer to the question completely at all. I would say two things, however, one goes back to what we did in the study in relation to the specific application in reinforcement and so on. One of the reasons why that was so important is because, I mean, this is, by the way, something very important about anthropological research that I didn't mention earlier is this commitment to wholism, right? You can chop life up into little compartments, you'll allow life to express itself in the messiness that it is, right? And that's how you gain insight, you know. If conversation wasn't allowed to go to questions of promotion, we would never have understood how important reports are in the context of people's trajectory, etc. So this holistic kind of approach, right? So you've got to look at what you're looking at reflexively, but also holistically, right? Now, take the reports as an example, but it's only an example. For as long as these reports are operative and catalytic, tomorrow and legal processes such as criminal law proceedings or complaints proceedings, where the attribution of blame upon humans is what's at stake, right? The whole design of the AI tool that enters into this process has to be oriented towards that. So basically, if I was talking about before that the new kind of social and legal technologies that might emerge out of this new world that we're in, if we arrive in 20 years in a world in which the hybrid or the network to use the tools of technology between AI tool and human is considered legally responsible. Then that's a game changer as far as how we think about that relationship. But for as long as the moral and legal buck stops with people as in flesh and blood people that you and me, right, and not with machines or the corporations that produce them, I mean, that complicates the matter of course. Then the way that we conceptualize the relationship in its practice will be slanted in that direction. So I think it's by taking that broader social historical context into account that you understand what it is that you're dealing with. If you don't, if you just look at this as a technical question, we do, you're missing the point, right? You need to look at it in relation to it. And the second point that I would make is just to maybe your listeners might find this useful. The kind of one or one example that we always use when teaching Bruno Latours idea of actor network through it was is the gun debate in the United States that might resonate. So that in the gun debate in the States, you got two camps, right? On one camp, we should get rid of guns because guns kill people, right? It allows people to do horrible things. And on the other hand, no, we should hold on to guns because guns don't kill people, people kill people, right? So it's gun versus person, right? Thing in the position of gun, you could put AI, right? Versus person. This is an ontological distinction, two different kinds of beings are at stake. And the debate about gun laws is passed out in relation to that ontological distinction. Now, Bruno Latours flat ontology says, guys, you got it all wrong because what's going on here is a person with a gun. That is the moral agent. It's the hybrid or networked entity or actant as he calls it technically doesn't matter. That's what's doing the killing, right? And that's where moral attribution is to lies, right? Now Latours took this really, really far. He wrote a book called Parliament of Things because the natural consequence of that is to incorporate inanimate non human objects into political proceedings, legal proceedings, giving them agency like you would to a person, which is what we're talking about, right? So I would transpose this to the AI debate and say, we need to experiment and think about what person with AI, right? Person with chat GPT. What kind of entities that, again, an ontological question? What difference ontologically does it make in terms of the potentials, the responsibilities, the biases, the trust, all the things that you talk about. We start thinking of people or employees and so on, or police officers as cyborgs to use a term of a good friend of Bruno Latours called Donna Harroway who works in the West Coast of the United States. She came out with this kind of cyborg manifesto. Taking seriously the hybrids between people and machines as moral agents in their own right, right? And not trying to break it down to ontological distinctions that are no longer applicable. Where does a machine end in a person's thought, right? That's why she used the metaphor of the cyborg. So I think that that's where it's at conceptually. Very interesting. What you've probably heard is that anthropology is not just a lens for studying the world as it is. It's a method of imagining how things could be. It's not just descriptive, it's ontological. In other words, it helps us rethink the various assumptions our technologies are built on. From how we define concepts like trust, fairness, or accountability, to how we understand the work of public safety itself. At Motorola Solutions, we often ask how to build better tools. But anthropology shows us that before you ask what to build, you need to ask what kind of world you're building it for. And what concepts you're bringing into that process. And that's why the kind of deep, grounded, human-centered research that Martin and his team led for police incident reporting and AI is so vital. Their work revealed that report writing isn't just data entry. It's a ritual, a form of professional performance. In AI doesn't just change the speed of that process, it changes the meaning of the report itself, how officers recall events, how and how narratives are assembled for courtrooms and communities alike. Anthropology helps us see those deeper stakes and navigate them with more care. In part two, we'll pick up where we left off, digging into a particularly powerful framework, actor network theory. We'll explore how it helps us understand agency as something that distributed across people, tools, policies, and environments, and how it can guide us in designing for messy networked reality of public safety today. Thanks for listening, and we'll see you in part two. [MUSIC]