Using examples from an ethnographic survey I conducted with manufacturers, researchers and users of so-called intelligent systems, this paper seeks to show that the relationship between humans and machines raises at least three categories of problems. The first one refers to their status and to the way they rely on life processes to function. The second emerges from the very particular interactions humans can have with them. Finally, the third one is related to the ways in which they reconfigure human organizations and activities. Leaning on the description of these ontological, interactional and organizational dimensions, this article will argue that we cannot fully grasp the effects of robots and AI without taking these dimensions into account. … [please read below the rest of the article].

Image credit: Sandra Strait via Flickr / Creative Commons

The problems associated with automation are certainly not recent. But today, as so-called intelligent autonomous systems integrate more and more domains of social life, these issues find new expressions. At the first glance these objects raise mainly technical problems. But they also arouse many hopes and concerns related to their power to transform society. What kind of society is being created around robots and artificial intelligence?

A World of Machines

Ethnography has given me an access to different ways of conceiving the future of our societies. It has allowed me to observe some of the techniques that are already used to transform them. Conducted in places where machines are imagined, designed and tested, my investigation has sought to better understand how robotic and intelligent systems participate in a global transformation which renew the ideas we have about life, social relations or even territories.

This comprehensive ethnographic approach seeks to overcome some of the difficulties that social anthropology faces in contributing to the study of the transformations associated with mechanical engineering and computer science. Of course studies have already been conducted. Stefan Helmreich has shown how computer simulation of life have an effect on its very definition (Helmreich 1998). Lucy Suchman’s (2007) work on the socio-material agency of human-machine interactions has already shown how difficult it is for humans to communicate naturally with machines. Other works carried out in France (Vidal 2007; Becker 2011; Grimaud and Paré 2011) also addressed the uncanny sort of communication with so-called intelligent machines.

As they now integrate an ever-increasing number of domains of social life robots and AI are raising a growing interest among social scientists. However, though intelligent machines share many features of a total social fact (Mauss 1925) the anthropological approach to AI is still searching its unity at a heuristic level. This article will try to bring a contribution on that particular aspect.

Today’s robots and AI don’t exactly look like the machines that engineers promised to us years ago. We are not about to live with friendly human-like slaves yet. Although anthropomorphic machines still capture the attention of most people (especially among anthropologists) robotics and computer science are no longer content to merely imitate human or animal bodies and activities. They do, as we are about to see. But they also seek to model many types of phenomena related to the simulation of ecosystems, economic and professional know-how, or the regulation of flows at very different scales while trying to implement these models in many sort of objects regardless of their shape. While they convey many scientific and cultural representations about the living, intelligent machines also give shape to a new type of relationship with our objects, but also to new forms of organization (Becker 2021).

Manufacturing Life-like Objects

Despite these machines appear to be very new, designing life-like objects has a long tradition. The defecating duck conceived by Jacques Vaucanson is one of the best known examples. This object does not only imitate the movement of a bird “cleaning its wings and feathers”. Its purpose is also to illustrate digestion. It is a machine that Vaucanson wishes “to be able to eat, drink, digest and empty itself” (Vaucanson 1738). The automata of the Classical Age represent the functions they imitate. It is thus not surprising that they played a crucial role in the development of science (Bredekamp 1995).

Today’s machines offer an extension to the automata. They tell a natural history which simulate and dramatize life by extending its limits (Riskin 2003). Like them, they constitute experimental forms of abstractions about the living (Helmreich 2011) and they are part of a much more larger attempt to domesticate vital processes. Indeed, many human groups use technical means to gain control over these processes, whether they consist in stabilizing their representations into objects (Coupaye and Pitrou 2018), in seeking to extend the limits of human action (Héritier 2007), or in conceiving the very humanity of humans (Godelier and Panoff 1998). How do roboticists and computer scientists address life? Let’s take the example of the Scratch robot.

Scratch is a research platform designed from the observation of rats. It is made of a body that carries a computer, a power supply, motors and wheels. It is also equipped with a mobile neck which control a triangular head. The head itself is equipped with whiskers moving back and forth. At the tip of the head, a muzzle is formed by a tactile sensor which is not articulated.

ScratchBot was born from a collaborative project funded by the European Union. The project aimed to artificially reproduce the sensory-motor abilities of rats in order to design a new kind of sensor. Since whiskers play a key role in the relationship rats have with their environment, mimicking this ability could ease the navigation of a robot, for instance in dark environments. But this research does not only aim at conceiving a new type of sensor. Its purpose is also to reverse-engineer the neuronal circuit stimulated by the sensations and the affordances provided when a rat explores its environment. The project is thus articulated around several disciplines such as psychology, neuroscience and engineering.

Of course, this machine is not a faithful imitation of a rat. It is not its purpose. It rather demonstrates a new method to detect obstacles automatically. The ambition of the project is to tackle existing problems within the fields of artificial perception, orientation, localization, cartography, representation and classification. But this machine also has a retroactive effect on knowledge.

Based on the principle that cognition relies on a biological circuit, the robot depends on a recursive approach which constructs an analogy between cognitive processes that are already considered as a technical object and a robot which functioning depends on an electrical circuit that imitate the biological one. This model helps to test the hypothesis that the superior colliculus is the organ determining the relationship between the stimulus and the behavior of the rat.

Such a bio-inspired object has also an aesthetic effect. Its activity gives the impression that it is alive. When the robot starts moving and reacts to obstacles, an observer will immediately wonder what principles lead its actions. Such a machine therefore encourages us to ask how autonomous objects are perceived by those who interact with them. How can such objects lead human interactants to imagine that they have a sort of life of their own?

“What Is That Thing?”

This aspect is perhaps the one that intrigues anthropologists the most. During their fieldwork, anthropologists are often confronted to the agency of objects. Intelligent machines do not escape the criteria which, in other contexts, allow anthropologists to describe the very particular mediations created by these objects.

For example, Denis Vidal has noted that robots provoke interactions which can easily be compared to those that humans have with objects in ritual settings. These objects engage people to interpret their behavior. They are the source of various beliefs and are at the center of debates and speculations about the principles that animate them (Vidal 2007). Despite their possible resemblance to a human or an animal, we are certainly never deceived about the artificial origin of machines’ movement. But at the same time, we are inclined to treat them as people. Experiments conducted at the quai Branly Museum with the robot Berenson illustrate this trend.

Born after the meeting of the neuro-cyberneticist Philippe Gaussier and the anthropologist Denis Vidal in the early 2010s, the Berenson project is an experiment in the wild aiming on the one hand to model the social learning mechanisms of aesthetics and, on the other, to better grasp the magnitude of the relations visitors have with an anthropomorphic robot thanks to video.

Interacting with the Berenson robot is definitely not an ordinary interaction. Once passed the uncanny feeling which characterizes most of the humans-humanoids encounters (Mori 1970), we can see that the status of the robot is subject to constant transformation and accumulation. The visitors’ comments cumulate criteria used to address a person and those used to describe objects. Berenson appear as a black box. And this implies to constantly requalify its status in order to verify that the machine is indeed one.

This test gives rise to strange formulations on the part of the visitors, which could easily be understood as direct attributions of intentions toward the robot. But they rather consist of game where projecting social qualities onto it helps to better define its mechanical nature. Does speech influence the robot’s behavior? What happens when we stand or wave the hand in front of it? Will it see us?

Visitors immediately recognize that Berenson is a robot. But they seek to gain knowledge about what it does by engaging in a relationship that takes the form of a social relationship. The presence of the robot is thus an opportunity to replay the ways in which social bonds are manifested; to rehearse some routines. It is common, for example, to see adults asking their child to address the machine politely. The presence of the machine is also an opportunity to break with the ordinary rules of interaction. The robot is often the subject of jokes, visitors call it loudly, stand in its way or wave their hands in front of its eyes. In some cases groups of children mock its strange appearance or event adopt a quite aggressive attitude toward it.

The work carried out on the Berenson robot shows how our relationship to so-called intelligent objects is transformed because of the uncertain situation they create. It also reveals a particular type of communication that deals with the very possibility to communicate. (Bateson 1972). An object like Berenson participates in a reconfiguration of the relations we have with our objects (Suchman 2007). However, these reconfigurations operate on a wider scale.

Machines We Live In

Paro gives a first example. A Japanese robot that looks like a baby seal, Paro is now finding a place in retirement homes. While its effects are difficult to assess, recent works show that robots can mitigate anxiety and loneliness (Robinson et al. 2013). Other researches confirm this trend.

A study conducted with the remotely controlled teddy bear Huggable and children hospitalized at the Massachusetts General Hospital showed that they were more likely to engage emotionally with a robot, illustrating the potential of social robots to provide support in mediated pediatric care situations (Jeong et al. 2015). In France, robotic mediation is receiving an increasing attention, which is principally motivated by a desire for caregivers to not leave the initiative to mere engineers in term of care protocols (Lafont 2021). This particular use of so-called intelligent machines shows that they are not necessarily intended to replace human at work. They can also constitute new therapeutic tools in a hybrid environment where humans and objects act together. These sociotechnical systems can actually take many shapes and raise particular questions.

For example, they can consist of regulation tools for transport flows. While expert systems have long been used to supervise road traffic, recent developments in AI suggest that they will undergo major changes. Motivated by various concerns ranging from safety to the optimization of supply chains, the design of autonomous vehicles combined with geolocation data flows or the implementation of the Intelligent Transportation System raise questions related for instance to the transformation of the existing infrastructure in countries that have already specific problems to solve regarding their own territories.

They also consist of automated production systems which redistribute the roles by mediating relations. This is particularly the case in dairy production. While robotic milking systems are an additional step in the rationalization of milk production processes, they reconfigure the work of the farmer and his relation to the herd. Implementing such a digital layer involves a deep transformation of the existing architecture. But the reconfiguration does not only concern the material aspects of the work environment. These systems also imply new arrangements between humans, animals and machines, which consist of a new attempt to synchronize biological and mechanical clocks in order to achieve an optimized level of performance.

These agricultural production systems can also aim for complete autonomy and therefore exclude humans from their operation. A research conducted in the field of space agriculture at the University of Arizona, the Lunar Greenhouse consists of a cylindric inflatable life support system which could one day be launched to the Moon or Mars. The module is conceived to provide edible vegetables, clean water and oxygen before the astronauts land. The balance of the system depends on sensors which data are processed by an onboard computer which can take action. Once they arrive, the astronauts would have thus to perform mainly maintenance tasks without necessarily having skills in agriculture.

By reconfiguring a community gathering humans, technical devices, animal or plants, these systems have an effect not only on the socio-material organization that makes them possible, but also on the organization of activities. Aiming at social homeostasis, they imply the implementation of new infrastructures which mediation also implies a reconfiguration of relations and activities.

Anticipations

Adopting the perspective of social anthropology, this paper aimed to identify the fields to be studied in order to address the effects that so-called intelligent machines can have for societies. It has tried to show that it is not possible to fully grasp the social transformations at work without considering at least three categories of problems.

The first one is ontological. It concerns the ambiguous status of objects imitating vital and cognitive processes which existence affects the representations we have of life. The second, which concerns interaction, addresses the particular relations we have with objects. Finally, the third one focuses on the way in which the increasing hybridization of sociotechnical systems participate in a reconfiguration of the community which affects not only activities but also their organization.

These problems are in direct continuity with the technical transformations associated to modernity. Vaucanson’s duck already addresses the role analogy can play in the understanding of life. The chess-player of Kempelen, despite its inauthentic character, already questions the relationship we have with life-like objects. And finally, the optimization of industrial production have questioned the organization of human activities since at least the first industrial revolution.

However, the development of electronic processes has eased our access to techniques employed by robotics and computer science. Though we are not about to live with C3PO yet, we now use so-called smart and connected objects on a daily basis. It is therefore urgent to acquire the tools to better anticipate their effects.

References

Bateson, Gregory. 1980 [1972]. “Problèmes de Communication Chez les Cétacés et Autres Mammifères.” Vers une Écologie de L’Esprit, t. 2, 137-154: Paris: Seuil.

Becker, Joffrey. 2021. “Vivre Avec des Robots / Vivre Dans des Robots: La Mécanique Du Lien Social.” In Jeunes Robots et Vieilles Personnes: Prendre Soin et Nouvelles Technologies en Gérontologie edited by Jérôme Pellissier, Mireille Trouilloud, Pascal Menecier, 29-58. Lyon: Chronique sociale.

Becker, Joffrey. 2011. “Récursions Chimériques: De L’Anthropomorphisme des Robots Autonomes à L’ambiguïté de L’image du Corps Humain.” Gradhiva 13: 112-129.

Bredekamp, Horst. 1995 [1993]. The Lure of Antiquity and the Cult of the Machine. Princeton: Markus Wiener Publishers.

Coupaye, Ludovic, and Perig Pitrou. 2018. “The Interweaving of Vital and Technical Processes in Oceania.” Oceania 88 (1): 2-12.

Godelier, Maurice, and Michel Panoff, eds. 1998. La Production du Corps. Paris: Éditions des Archives Contemporaines.

Grimaud, Emmanuel, and Zaven Paré. 2011. Le Jour où les Robots Mangeront des Pommes, Conversations Avec un Geminoïd. Paris: Éditions Pétra.

Helmreich, Stefan. 2011. “What Was Life? Answers from Three Limit Biologies.” Critical Inquiry 37 (4): 671-696.

Helmreich, Stefan. 1998. Silicon Second Nature, Culturing Artificial Life in a Digital World. Oakland: University of California Press.

Héritier, Françoise. 2007. “Chimères, Artifices et Imagination.” In L’Homme Artificiel, Colloque Annuel edited by Jean-Pierre Changeux, 39-59. Paris: Odile Jacob.

Jeong, Sooyeon, Deirdre Logan, Matthew Goodwin, Suzanne Graca, Brianna O’Connell, Honey Goodenough, Laurel Anderson et al. 2015. “A Social Robot to Mitigate Stress, Anxiety, and Pain in Hospital Pediatric Care.” HRI’15 Extended Abstracts. Portland, OR: 103-104.

Lafont, Véronique. 2021. “Expérience et Questionnements sur L’Utilisation du Robot Paro dans un Ehpad.” In Jeunes Robots et Vieilles Personnes: Prendre Soin et Nouvelles Technologies En Gérontologie edited by Jérôme Pellissier, Mireille Trouilloud, Pascal Menecier, 147-152. Lyon: Chronique sociale.

Mauss, Marcel. 2007 [1925]. Essai sur le Don. Forme et Raison de L’Echange dans les Sociétés Archaïques. Paris: PUF.

Mori, Masahiro. 1970. “Bukimi No Tani (The Uncanny Valley).” Energy 7 (4): 33-35.

Riskin, Jessica. 2003. “The Defecating Duck, or the Ambiguous Origins of Artificial Life.” Critical Inquiry 29 (4): 599-633.

Robinson Hayley, Bruce MacDonald, Ngaire Kerse and Elizabeth Broadbent. 2013. “The Psychosocial Effects of a Companion Robot: A Randomized Controlled Trial.” Journal of the American Medical Directors Association 14 (9): 661-667.

Suchman, Lucy. 2007. Human-Machine Reconfiguration, Plans and Situated Actions, 2^nd Edition. New-York: Cambridge University Press.

Vaucanson, Jacques. 1738. Le Mécanisme du Fluteur Automate Présenté à Messieurs de L’Académie Royale des Sciences. Paris: Jacques Guérin.

Vidal, Denis. 2007. “Anthropomorphism of Sub-Anthropomorphism? An Anthropological Approach to Gods and Robots.” Journal of the Royal Anthropological Institute 13 (4): 917-933.

Categories: Articles

Tags: anthropology, artificial intelligence, Berenson robot, Denis Vidal, digital transformation, human-machine interaction, Joffrey Becker, Paro, Philippe Gaussier, robots, ScratchBot