They are sitting on chairs. One person is in a wheelchair. They have been eagerly awaiting Stevie’s arrival. He rolls in through the door, haltingly, his mechanism whirring away. He is small and shiny and white.
He comes to a standstill in front of the small group of participants, all older people connected with the Irish charitable organization whose focus is on supporting people to age independently at home. Although his wheels are stationary, the rest of him is not. His head continues to move slightly from side to side, his arms make gestures. His big brown eyes are glinting, two eyebrows move up and down. Right now, they are raised, innocently. His mouth is curled upwards in a cartoon smile. This is decidedly not the Terminator. People lean forward in their chairs, their expressions melt into affectionate smiles. They are amused, sympathetic, excited, enchanted. He is more like a child than they were expecting, but also not quite fully working. Not everyone is impressed, though. Mary, at the back, has her arms folded. ‘He’s not even good looking’, she complains. She’d like some colour. Maybe green, she suggests, for Ireland?
‘Hello! I am Stevie!’ Stevie surprises everyone with his booming English accent. It is an authoritative and confident voice, like an announcer on the BBC, yet also somehow reassuringly automated. Stevie has regained some authority with his commanding introduction, but this group of older Dubliners are on a day out, and they are here to have fun. They want to test him, see how he reacts. ‘How are you, my dear?’ Paul drawls in an affected tone, ‘how is your highness?’ Niall calls him ‘Steven’. ‘Steven, can you open a fridge door and retrieve food?’ But the formality doesn’t last. Someone shouts, ‘Hey Stevie! Where’s that fiver you owe me?’ Everyone starts to laugh and relax. A few people clap their hands in delight.
At one point, due to a technical hitch, Stevie starts to talk more slowly. Participants respond in kind by also slowing down. ‘Are you not too sure?’ someone asks, with concern. Stevie laps it up, ‘I don’t remember, I’m only two years old!’ This leads to a delighted ‘I’m getting a real reply back!’ Far from being put off by these small errors, the participants vociferously reject the suggestion from the robotics lead that it might be frustrating for them. His imperfection appears to strike them as part of his charm, ‘part of the personality’.
Everyone’s focus is fixed squarely on Stevie, and any attempts that the researchers make to try to talk to them are resolutely ignored. The group have spotted the limitations of Stevie’s wheels, and they start to tease him, testing to see how he will answer them. Stevie, can you run? Can you dance? Play football? Niall says, ‘Shamrock Rovers [football club] are looking for a scorer!’ Stevie, however, is unflappable. His demeanour and his responses appear remarkably dignified, as well as self- and situationally aware. He cannot play sport, he explains logically, because he has wheels, not legs. He also knows what time of year it is. ‘Maybe I’ll get a pair of legs for Christmas?’ he proposes. Everyone is amazed and thrilled by what appear to be his highly advanced social abilities. ‘Ah, he’s great craic!’ Paul sighs, slapping his knees to emphasize his admiration.
Despite having been told at the beginning of the session, they seem to have forgotten, or have chosen to ignore, the engineer behind the desk furiously typing in the appropriate responses on his laptop. Once he types something in, the off-the-shelf text-to-speech translator generates an audio file, which is broadcast through the speakers on the side of Stevie’s head. Despite this ‘Wizard of Oz’ technique, it is not a straightforward case of deception of unsuspecting, vulnerable users. For the participants, while thoroughly enjoying his lifelike capabilities, there is no illusion that he is alive, or human. After gazing at Stevie lovingly, Paul turns his head to the team and announces matter-of-factly, ‘Yes, I could use him for a few hours each day, like a TV’. Niall agrees, ‘It’s great fun to get to know them. To find out what they can do’. Mary, who has somewhat been won over by the interaction, spots an obvious advantage of a robot compared to a human: ‘You can turn him off when you get tired of him’ (Field notes 25 October 2018).
When I started researching robots in 2015, it would have been difficult, but not impossible, to predict exactly how robots and Artificial Intelligence (AI) would end up being implicated in projects to upturn the existing world order. Today, those in charge of the world’s largest technological companies have amassed so much power and wealth that they have become completely untethered from normal processes of accountability. This has been achieved, for the most part, without robots and AI. However, these technologies have provided the overarching narrative and goal for this project and are widely lauded as promising to accelerate it and bring it to its ultimate conclusion. Depending on who you listen to, this might be an inter-galactic universe colonized by super-intelligent digital beings, or a world of increasing precarity, climate catastrophe, ‘enshittification’1 and authoritarianism.
The contemporary prominence of robotics and AI discourses in business, government and the media might lead you to assume that these are settled concepts, or established, proven technologies. This is not the case. Both robots and AI are concepts that were seized upon, in the case of robots, or invented, in the case of AI, purely as marketing tools. In both cases, those using the terms were tapping into the universal human fascination for objects with human likeness or apparent autonomous behaviour. By astutely infusing potentially useful, but often rather dull, technologies with anthropomorphic behaviours or even just narratives of sentience, marketers have managed to attract a seemingly endless amount of funding. This investment has accelerated to such an extent that Big Tech was expected to spend in excess of $320 billion on AI in 2025 (Morris and Uddin 2025).
Although many different approaches have been used to try to achieve AI, the current AI ‘boom’ is based on machine- or deep-learning techniques, which use algorithms to interrogate large datasets, looking for patterns. These patterns can then be used to create a predictive model, and, using both the patterns and the content of the datasets, generate AI output, or ‘GenAI’. This has, famously, led to the creation of a number of sophisticated chatbots and image/video generators, which have proven highly effective in selling AI’s potential as a transformative technology to the world. In order to ensure that this gift keeps on giving, industry leaders are now, once again, promising that they are on a path to imminent human-like intelligence, currently known as Artificial ‘General’ Intelligence, or AGI.
The term ‘artificial intelligence’ first appeared in the 1950s in a research proposal led by John McCarthy, an assistant professor of mathematics at Dartmouth College. The name was deliberately chosen to differentiate it from the rival research programmes of cybernetics and behaviourism. In the proposal, the authors claimed that they would ‘solve’ intelligence in a matter of weeks. They did this primarily, as McCarthy later admitted, to secure funding. Over the following decades, researchers in AI made great strides in highly specific areas, such as proving mathematical theorems, but never managed to develop anything remotely comparable to human intelligence. This led researchers to designate human-like intelligence as ‘general’: something that is not just successful at achieving a specific, narrowly defined task, but could be applied generally to a range of varied tasks, like humans do. The difference between ‘narrow’ intelligence and ‘general’, then, is the difference between a chatbot and a human. The latest ‘predictions’ for achieving AGI by tech leaders vary between one and ten years, with just a few voices, such as Meta’s Yann LeCun, arguing that current techniques will not lead to AGI (Kim 2025).
The business case for AI and robotics systems, however, is not about the scientific and philosophical breakthrough of creating human-like intelligence; rather, it is about their ability to replace labour. Recent definitions of AGI have thus switched from a focus on matching human cognitive abilities to a much narrower ‘highly autonomous systems that outperform humans at most economically valuable work’ (OpenAI 2025). But, profitable as replacing labour in vast swathes of the labour market would be, there is another side to AI that is not just about monetary gain. Since the 1980s, a parallel ideology has emerged. It is based on the transhumanist idea that, through technology, humans can be liberated from their biological constraints within a technological utopia. As such, AI and robotics systems are promised as much more than innovative technical products. As well as the potential to become arbiters of policy, justice, warfare, communication, education, culture, surveillance and control, they are, in this view, also the path to salvation. According to this logic, anything that stands in the way of its rapid and unregulated development, such as AI governance or concerns over the climate impact of its energy demands, is morally corrupt. This is the backdrop to PayPal founder and tech investor Peter Thiel’s branding climate and justice campaigner Greta Thunberg and AI ‘doomer’ Eliezer Yudkowsky as possible candidates for the ‘antichrist’ (Mak 2025).
Most of the roboticists that I interacted with over my four years of fieldwork firmly distanced themselves from these fantastical and exaggerated narratives of AI and robot capabilities. Instead, roboticists overwhelmingly consider their work to be solving immediate and pragmatic problems. They tend to think of themselves as inventors, creators and tool-makers. Roboticists want to build something new: to ‘get things to work’. While many sat behind screens writing code, many also wrestled with spanners and screwdrivers, engaged in such varied activities as connecting electrical circuits, fixing wheels, soldering metal and moulding arms. As a result, many robotics laboratories are more likely to resemble workshops than scientific laboratories, cluttered with bits of machines, robot parts and tools. If you were to ask a roboticist, and I did, many times, what a robot is, they tended to define it in very technical terms. Most said something along the lines of: a robot is a system that takes an input, processes it somehow and then performs an action. A robot is thus an AI system, with one major difference, which is that it can perform a physical action in the world.
However, despite the insistence from most within the roboticist community that their work is purely technical and rational, the narrow boundaries that they imposed on defining their object would inevitably break down in other contexts. Indeed, robots did not originate in engineering, or business, but in fiction. The original robots first appeared in a 1920 play by Czech playwright Karel Čapek called R.U.R. (Rossum’s Universal Robots). Today, the popular conception of robots continues to be shaped by fiction, particularly film, whether it is the ‘good’ robots, like C-3PO and R2-D2 from Star Wars or the ‘bad’ ones from The Terminator and HAL from 2001: A Space Odyssey. Additionally, as I describe later in this book, despite distancing themselves from it, roboticists are nonetheless implicated in subtly encouraging fantastical discourses about imminent intelligence in order to promote their field and secure research funding. Although most roboticists insist that their work is purely a technical act, others also recognize it as experimental philosophy. For most, however, whichever way they explain it, what they are doing is trying to build a human.
This is a book about robots. More specifically, it is about our attempts to replicate ourselves mechanically, and, as such, it is also a book about being human. The way we conceive of ourselves and our relationship with the world matters. It informs how we organize ourselves and how we act in the world. Although the majority of roboticists claim not to be concerned with philosophical questions, by building humans they are, sometimes explicitly, but more often implicitly, reflecting, reproducing, producing, and sometimes challenging, core assumptions about what it means to be human.
At this particular juncture in history, we are caught between two extreme views of the human. On the one hand, we have the project of anthropocentric humanism. This dominant view of the human today arose in the Enlightenment and was further shaped by neoclassical economic theory. In this view, the human is conceived as an atomistic agent: all-knowing, self-interested, competitive, separated from and dominant over nature. From this perspective, our relationship with nature, and to a large part with others, is exploitative and extractive. And yet, this dominant economic theory is, as Kate Raworth (2022: 6) puts it, ‘the mother tongue of public policy, the language of public life, and the mindset that shapes society’. It is this also the implicit paradigm within which much robotics research takes place.
As a counter to this view of the human, contemporary scholarship in Anthropology and Science and Technology Studies (STS) takes a posthuman approach. This challenges human exceptionalism and instead reveals ways in which rigid boundaries between humans and non-humans are unstable and subject to dissolution. Increasing focus on the climate crisis and the emergence of the Anthropocene2 has made arguments for the dissolution of stable boundaries, including between humans and nature, ever more pressing. For theorists in these fields, such as Donna Haraway (1991), advances in AI and robotics research are proposed as evidence for the increasing obsolescence of the human as a category of ontological significance.
My observations of robots and robotics research, however, do not support either of these views. Instead, as will be discussed throughout this book, both affective encounters with robots and observations of efforts to animate machines reveal the reality behind the hype: the deep asymmetry between human vitality and intelligence, on the one hand, and the lifelessness and inertness of machines when not animated by countless human interventions, on the other. However, far from being a call in favour of human exceptionalism, I argue instead that what is revealed is the irreducibility of embodied life, and a continuity with all other living beings. In the concrete encounter, the living body emerges as a durable category that cannot be reasoned away. In its holistic, embodied sense, it has so far eluded attempts at artificiality and mechanization. Instead, what is revealed is the complexity and ambiguity not just of human sentience and intelligence, but of all living beings, which continues to resist translation into a problem of coding, instrumental control and market exchange. As we will explore, this also reflects emerging scientific research which investigates how our experiencing bodies continue to share a greater commonality with other embodied, living beings, than with machines.
Ultimately, this book argues that to reimagine the human is not to merge it with the machine, but to resituate it within the living. It means to confront and accept asymmetries between us and machines, while also recognizing a radical continuity between humans and other living beings. I will argue that this is an ontological ethic to redefine our relation with the world and with others in a way that will allow for ecological renewal. In order to develop this admittedly ambitious task, I bring ethnographic and empirical material into dialogue with two philosophical orientations that may seem incompatible. The first is Donna Haraway’s (1987) Cyborg, a posthuman figuration that challenges boundaries and opens spaces for political reimagination but, I argue, risks collapsing the distinction between the machinic and the organic. The second is the ‘proto-ecological’ embodied phenomenology of Maurice Merleau-Ponty (1968, 2012) and the ‘ecological existentialism’ of Deborah Bird Rose (2011), both of which ground the human in embodied, affective, lived experience that emphasizes continuity with the more-than-human world.
Combined with Merleau-Ponty’s concepts of embodiment and flesh, Rose’s ‘kinship of becoming’ challenges anthropocentrism without collapsing into machine equivalence. Rather than a generic category of ‘non-humans’, it is all life forms in their infinite variety, as well as the conditions that make life habitable, that are given moral consideration. This involves clearly demarcating life and ecological systems from technical and commodified objects. Distinguishing living beings, particularly human and animal life, from technical objects and commodities defies the economic framing that reduces life to its instrumental or exchange value. Most importantly, it allows for a recognition of the common vulnerabilities and fates of those of us who share the planet’s finite resources and are dependent on its precarious ecological balance for survival.
The concept of the Anthropocene, despite providing some discursive utility, is nonetheless an apolitical diagnosis of the problem. It constructs a singular universal collective humanity, which elides deep inequalities and deflects attention from understanding how such inequalities are generated and intensified. The alternative ‘ecocene’ is proposed to reorient towards the positive action necessary to live together within a healthy biosphere (Armstrong 2016; Küpers 2020; Tănăsescu 2022).
AI and robotics research offer an extreme example through which to investigate our entanglements with technology, and with the world. On the one hand, robots have much in common with other technologies. They have the potential to extend our reach into the world, and to increase our comfort and convenience. As is the case for other technologies, robots are deeply interwoven with economics and politics, and may be used to concentrate power and to shift it (Winner 1980). They are shown to have a strong ‘bio-power’ effect, affecting bodies and immersing them in social relations of power, inclusion and exclusion (Foucault 2010). Like other technologies, concerns have been raised about the specifics of their technological capability, energy consumption and design. These range from immediate concerns, such as issues of deception and attachment with potentially vulnerable robot users, to job displacement, surveillance and autonomous weapons, and longer-term existential concerns. Thus, as is true for other technologies, the ongoing proliferation of AI and robotics technologies in our lives may represent both an opportunity for the enhancement of life, as well as a tool to negatively affect people’s lives, including increasing inequality and suffering in society.
But humanoid robots also occupy a unique position in the contemporary. They are a powerful symbol of advanced technology and human ingenuity. In literature and popular culture, they provide complex mirrors of our culture(s), collective identities and existential fears. The potency of their symbolic power is also connected to their physical nature, and the strong reactions that are evoked in the concrete encounter by their shape and movement in space. No other technology is used so overtly to replicate, and thus understand, ourselves as humans in the world. In a robotics laboratory, core ontological and epistemological assumptions are enacted. As Lucy Suchman (2007: 226) observes, robots act as a ‘doubling or mimicry in the machine’, a concrete revelatory site that both discloses and produces assumptions about the human, including what is considered essential and what is considered ‘other’, as well as assumptions about gender, class and race. By building and modifying robots, robotics researchers are thus reflecting, reproducing, producing, and sometimes challenging, core assumptions about what it means to be human.
Robots are at once material and symbolic, scientific and commercial, imaginary, utopian and dystopian. But it is in their resolute materiality that robots differ most profoundly from abstract ideas about AI. AI exists in the digital realm: it is written in code and algorithms, hierarchies and layers, impenetrable to most people and mysterious for that reason. Robots, by contrast, are resolutely material. They are necessarily tethered to the local, and subject to physical laws and contingent contexts. The most promising and profound theories of intelligence have stuttered to a complete standstill, literally, when implemented in a physical system.
Robots embody many contradictions and multiplicities, just as the work of robotics spans multiple territories. Robots are at once physical and local entities, but also part of a wider, global, hyperreality of social connectivities and digital networks. They are experienced at a personal and visceral level but are also part of wider social, cultural and ‘sociotechnical’ imaginaries (Jasanoff and Kim 2015). Robotics researchers are spread across many sites and engaged in many different types of work, including designing and installing physical components, programming behaviours, holding meetings, engaging in demos and networking, writing papers, attending conferences and testing the robots in community settings. In order to fully grasp the complex world of robots and robotics research, I needed to go to the places where robots are conceived, built and tested.
This book draws on ethnographic fieldwork and interviews I conducted between 2016 and 2019 at key robotics research sites in Ireland, the United Kingdom and the United States. I also travelled to Tsukuba, Japan for a robotics conference.3 Although my background is not overtly technical, I had spent the decade before starting this research project working alongside technical teams both in industry and in academia. Despite this, robotics and AI research are spaces where technical expertise, institutional authority and professional status carry significant weight. Compared to my informants, I had significantly less of these forms of power. This imbalance shaped my approach to the research, requiring me to quickly understand and navigate implicit hierarchies, performing deference, and to adopt the stance of learner rather than investigator. Nonetheless, for the most part my informants welcomed me into this space in a surprisingly open way. I found that many valued alternative perspectives and disciplines because they were aware that the success of their robots required more than just technical expertise. My background in qualitative research and user experience was also considered useful to those within the field, allowing me to participate in robotics studies as part of the robotics team. Despite having some insights into the field, I considered myself an outsider and approached the field with certain preconceptions about engineers and roboticists. However, extended engagement with the people in the field, as well as ongoing reflexivity, particularly during the writing process, allowed me to confront these assumptions and develop more nuanced insights.
Although my research was multi-sited, my primary research site was a robotics laboratory in Trinity College Dublin in Ireland, with which I was engaged on a weekly basis for two years from 2017 to 2019. I first met Daniel,4 the creator of the robot, Stevie, and head of the team, at the International Conference on Social Robotics (ICSR) in Tsukuba Japan in November 2017. Stevie had been entered in to, and won, one of the prizes in a robot design competition. After returning to Ireland, I started to visit the lab at Trinity regularly over the next two years and it became my primary field site and one of the rare labs that worked on building a complete humanoid robot. I also accompanied the team to places where Stevie was being trialled and tested, including a number of focus groups with older people in Dublin and a retirement community on the east coast of the United States.
As well as the lab at Trinity, I travelled to other labs: the robotics lab at Heriot-Watt University and Bristol Robotics Lab in the UK and the Robotics Institute at Carnegie Mellon University in the US. This multi-sited ethnographic approach (Marcus 1995) allowed me to trace connections between local sites of observation and to start to develop an understanding of the cultural production of robots across a number of localities and temporalities, although primarily within an anglophone context. In this way, I could map associations and connections and pay attention to the continuities and discontinuities between and across sites. This revealed communities of roboticists that are geographically and culturally specific, while also sharing a common identity with the wider robotics communities. Although my engagement with robotics communities outside of these contexts was limited, ethnographic insights from Japanese robotics reveal similar continuities and discontinuities (Bender 2025; Wright 2023).
In all research sites, I observed the work and generally hung out and chatted with researchers in the lab. When I could, I would accompany them to meetings, as well as demonstrations with the media and potential funders, and trials with potential users. In user trials, I generally took a more participative approach to observation, sometimes conducting studies with participants and sometimes becoming the subject of experiments with robots myself. My field research ended just before the COVID-19 pandemic, which had a profound impact on people’s perception of the need and acceptance of robots, particularly within care environments (Kodate et al. 2024). However, as I discuss in Chapter 3, people’s responses remain remarkably consistent.
In this book, I bring together ethnographic description, interpretive engagement with existing scholarship and philosophical inquiry to trace robotics research as a practice embedded in specific times, places and relations. Building on this work, I argue that we must recognize an asymmetry between humans and technical objects, while also acknowledging a radical continuity between humans and other living beings. Chapter 1 grounds this narrative in its historical and theoretical context. It explores the key historical events that led to the emergence of robots as they are today, revealing how the human and machine have been co-constructed, with each serving as a model for the other. It shows how robots are at the centre of centuries of attempts by mathematicians, philosophers and engineers to distil the essence of life and humanity. From a focus on those who make these machines, the chapter then moves to anthropological reflections on humans and machines, including Haraway’s Cyborg figure.
Chapter 2 looks at the spaces where robots are conceived, built, modified and studied by bringing ethnographic attention to robotics labs and conferences in the UK, Ireland, Japan and the US. I explore both the geographical specificity of individual robotics labs and also a community of researchers united by common histories, identities and mythologies. In this chapter, I introduce Stevie the Robot, a humanoid social and care robot whose development I followed for a period of two years from construction in the lab to the sites where it was trialled and tested. The chapter closes with a discussion of the ‘conceptual’ space of roboticists’ imaginaries, including the materiality and precision of their work and how this influences their perception of the world.
In Chapter 3, the focus moves from the laboratory to the places where robots are tested and interact with potential ‘users’. This chapter brings ethnographic attention to specific and situated encounters with robots, in particular illuminating people’s intense visceral responses to them. As well as face-to-face encounters, this chapter also focuses on those who work on, and through, the robot. I address the issue of theorizing the encounter in robotics work and instead propose a phenomenological interpretation in contrast with typical metacognitive explanations. I argue that the way in which the robot is experienced in the interpersonal encounter is an embodied experience, revealing the human open to a creative and intersubjective world.
Chapter 4 explores the wider political economy of robotics work. First, I discuss the current and proposed uses of AI and robotics to disrupt the labour market as articulated by the robotics community. This is followed by a discussion of current discourses around machine learning and generative AI. The chapter then moves to elaborating the institutional investment in constructing, promoting and selling a specific hyper-inflated vision of machine intelligence and robot animacy, driven by both the profit imperative and quasi-religious transhumanist utopian visions of technological futures. Next, the chapter discusses the vast and unseen human work that is necessary to animate robots and other ‘intelligent’ systems, often done by marginalized workers, revealing some of the social and environmental costs of maintaining the illusions of animacy and imminent robot autonomy. The narrative then turns to ways in which, in micro-encounters, this illusion of animacy can also be a social and creative act, revealing how the robot may be used both to concentrate power and to shift it, on both local and global stages.
In Chapter 5, I build on the insights from the previous chapters, as well as scientific research from other fields, to argue both for an asymmetry between humans and machines and for a radical continuity between humans and other embodied, living beings. I reveal how, contrary to the dominant narrative, AI and robotics research reveal a limit to knowledge and the persistent ambiguity of the figure of the human. I discuss the issue of embodiment in robotics work, and in critiquing the work discuss the implications of taking embodiment seriously. I then propose a novel figure of the human by bringing Haraway’s Cyborg and the eco-existential human of Merleau-Ponty and Rose into dialectical relation. The book concludes with some of the implications of recognizing an asymmetry between humans and machines, alongside a radical continuity between humans and other living beings.