Magic, Ritual, and Witchcraft

introduction

Over the past fifty years, robots have played an increasingly important role in human society: in healthcare, industry, scientific research, and the military. The widespread deployment of these machines is both result and partial cause of several technological "revolutions," revolutions that are significantly changing human attitudes, beliefs, and relations to their environment. Digital technologies are frequently referenced in religious contexts, with descriptions of these technologies as a new medium for expressing transcendence and spirituality. Additionally, robotic technologies are sometimes referred to as "haunted" machines. Philosopher Jacques Ellul was among the first to identify and analyze the progressive sacralization of technologies.¹ In this regard, technologies function not only as vectors or channels of the sacred in modern times, but also as objects upon which a certain sense of the sacred is projected and focused.

The debate around magic in / and high technology has however raised quite polarized views: some scholars, such as Hervé Fischer and Erik Davis, claim that digital technologies are replete with (a reinvented and modern form of) "magic" since they are source of wonder: on this view—the hypermodern technological imagination is stuffed full with mythical and magical allusions.² At the opposite pole, philosophers such as Rémi Sussan and anthropologists such as Emmanuel Grimauld have expressed critical views [End Page 480] on the alleged continuities between ancient forms of magic and modern allusion to it.³

These conflicting stances about magic in modern technologies do not preclude references to supernaturalism in the study of robots and robotics. Engineers even emerge as the "shamans" of modern and technological times; for Erik Davis, there are good reasons to take seriously into consideration recent references to magic in discourses on High Tech, AI, and robotics.⁴ Such discourses on "magic" in technological contexts does not by itself however corroborate the fashionable hypothesis that ancient belief systems, developed in traditional societies and ranged under the category of "magic," subsist as such in modern societies, a point that has notably been discussed in depth by historians of antiquity.⁵ Scholars in technology and robotics have been eager to borrow from anthropological theories of "primitive" magic,⁶ whereas anthropologists and historians, familiar with the symbols and practices of magic, have only recently considered extending the models of traditional magic to the hypermodern world of technology. It is as if the discipline specialized in the "primitive" was reluctant to extend the allegedly "primitive" category of magic to the modern world and to the digital technologies that characterize this modern world.⁷

The approaches mentioned above consider the sacred in relation to religion: in so doing, they rely upon an analogous relationship between technology and religion that prevails in the contemporary debates, since the relationship to technology is considered allusive in this case (when it's about "the religion of technology") but as realistic in other cases (when technology impacts religious behaviors). Still, another dimension of human beliefs nowadays resonates significantly in High Tech and robotics—that of magic.

Allusions to magic have indeed colonized public and scholarly debates on the digital revolution: this is the case for the study of the internet,⁸ and of [End Page 481] technologies augmented by means of digital devices.⁹ This scholarship unfolds against a backdrop of massive digitization of culture and society, with a resurgence of so-called "irrational" thinking, that has been particularly emphasized in the fields of marketing and economics, where references to theories of magic in anthropology are common.¹⁰ The media coverage of the massive digitalization surfacing in the mid-1990s made frequent reference to a "magic in the black box," a slogan that was already widespread in the media and public discourses on technologies.¹¹ In the modern context of high tech, these discursive references to magic have also become infatuated, especially in the case of AI.¹² Modern robotics, surprisingly, proves to also be a fertile ground for the reinvention of magic—at least under the form of mental attitudes (so-called "magical thinking"), if less so with reference to rituals and traditions (magic as / in cultic practice).

The main objective of this article is to assess the empirical pertinence of the concept of magic within a particular context, namely that of a robotics laboratory. The objective is not to align on a specific theoretical framework for magic, nor to depart from the a priori conceptual assumption that magic "exists" as such in robots and in the minds of those who create them. Instead, I advance an inductive approach, in search of empirical expressions, performative and discursive uses and meanings of magic, in a laboratory of robotics.

robots and belief: from magic to religion and back again

Robots are supposed to epitomize rationality. But as roboticists already know quite well, the story of robots traces back to a religious history, with references to Greek Mythology (Talos)¹³ or the Jewish Golem—two epitomes of "robots" before the industrial era.¹⁴ Considering such a genealogy, it is easy and common to support the existence of an ontological link between modern robots [End Page 482] designed in a secular context and the automata that preceded them in religious settings. After having been imagined in ancient mythology¹⁵ and developed in liturgical contexts in medieval Europe as automatons in churches,¹⁶ robots have recently been designed in "theomorphic" shape¹⁷ and liturgical services worldwide in temples: the Buddhist Mindar in Japan¹⁸ and Xian'er in China, the Catholic Santo and Celeste in Italy, the Protestant BlessU2 in Germany,¹⁹ and the Hindu Ganesh in India²⁰ are all remarkable examples of this "boom" of machines in the domain of the sacred. Yet, the emphasis has been put on the religious design, use, and meaning of articulated and moving machines in the longue durée, from ancient times to modernity,²¹ leaving almost unexplored the magical aspects of robots and the ways humans interact with them, especially since the Middle Ages.²² Christian Fron and Oliver Korn, however, have pointed out that while religion did play an essential role in the development of technologies and animated machines, a shift in representations is remarkable since the Middle Ages onwards, when automata and (future) robots came to be associated with ideas of sorcery, which may explain why they are still currently associated with the idea of "magic."²³ How relevant, however, can "magic" be for modern robots, since it has been primarily framed after the study of ritualism and supernaturalism in traditional societies? Not only does "magic" surface more and more in the narratives of [End Page 483] "technological enchantment" of modern societies,²⁴ but recent studies, like Musiał's monograph on "enchanted robots," have proved appropriate to refer to "magic" to study modern robots, from the point of view of empirical philosophy.²⁵ The anthropologist Denis Vidal even alluded to animism in robotics²⁶—a concept associated with magic since Edward Tylor.²⁷ Attributing intentions to machines is indeed part of the anthropomorphic projection known to anthropologists, more or less supposedly reflecting the continuity of traditional societies' animistic thinking in modern societies.²⁸ In Musiał's case, the study of technology calls for a theory of magic; in Vidal's case, an update of theories of magic should be engaged after the study of technologies.²⁹

Further, a first interesting conceptual contribution came from modern robotics itself in the form of the theoretical model elaborated by the Japanese scholar and engineer (and now faithful Buddhist) Masahiro Mori, who developed the theory of the "uncanny valley," in which he modeled the reactions that arise from relationships of humans with different forms of nonhuman entities, from dolls to robots.³⁰ This experience of 'strangeness' is assumed to nurture different kinds of feelings that bear similarities to magical and sorcery-like beliefs as they have been recorded throughout cultures from ancient history into modern times.³¹ Mori however made no explicit reference [End Page 484] to magic nor irrational beliefs; neither did he refer to any kind of supernaturalism, at least outside the tradition of Buddhism he admired.³² While anthropological theories of magic are extensively mentioned in robotics and AI,³³ the psychological mechanisms described by Mori equally, and reciprocally, infuse an anthropology of robots.³⁴

Whereas in the 2020s, experimental studies in high tech and robotics have started to record and analyze empirical expressions of emotion, cognitive processes, and psychological projections and characterize some of them as corresponding to "magical" thinking, few ethnographic studies have so far been conducted by scholars in social sciences and humanities on this topic in laboratories.³⁵ As an anthropologist of religion myself, magic and witchcraft range "naturally" among my areas of interest and specialization. Faithful to the anthropological perspective, any research on a topic such as magic must be based on an empirical understanding of what it means for people and how it is expressed in practice, through language and acts. A tendency of anthropology is the temptation to extend its theories on universal features of humankind, developed on the basis of studies of traditional societies, to hypermodern societies: this applies to rituals and sexual prohibitions; it could also be the case for magic, which seems particularly suited to beliefs based on causalities alternative to rational thought both in traditional societies³⁶ and hypermodern ones³⁷—albeit the hegemonic position of rationalism in Western worldviews leaves little room for other systems framed after magical thinking, considered as illegitimate, despite their long-standing presence in the history of cultures and societies.³⁸

Adopting a fresh perspective, i.e. considering that magic is a symbolic resource for modern contexts as much as it was for traditional ones, this article aims at exemplifying the way magical symbols, ideas, and behaviors can surface in the very heart of modern robotics, i.e. in a laboratory. The [End Page 485] empirical grounds for the paper are a case study examining the ways to identify, collect, and record expressions of magic in a context supposed to epitomize technological modernity and rationality. The article consists of an ethnographic evaluation of the relevance of the reference to magic, on the grounds of a series of preliminary ethnographic vignettes. The key contribution lies in the focus on empirical expressions of magic in the heart of a social and technological environment that have allegedly nothing to do with "irrational" or "sacred" ideas and beliefs. In other words, we will address the following question: if many observers of the massive digitization of modern societies have been inclined to support the idea that magic resides in technologies and robots, to what extent is this perspective still relevant with a close-up look? In other words, is robotics a fertile context for "magic," and if it is the case, what does "magic" stand for in the robotics lab? The empirical scope is here restricted to robots, and this is all the more challenging since such machines have been significantly more associated with religion than magic. Magic, however, plays a key role in the mythical imagination of robots, and their materialization under mechanical forms in society, yesterday and today.

toward an ethnography of magic in robotics: perspective and methodology

Contemporary discussion on the "sacredness" and the "magic" of machines remains broad: there has been little empirical evaluation of magic in robotics, as opposed to the allegoric magic of robots. Accordingly, this article aims to draft ethnographic foundations for an approach to magic in laboratories, which is less developed in anthropology than empirical investigations in others compartments of modern societies (hospitals, schools, and domestic spaces, for example).³⁹ Given the crucial position of magic in the agenda of anthropology, one must guard against a bias. Prudence must be the rule: anthropologists' familiarity with ambient magic in so-called "primitive" societies can indeed lead to mechanical identification of the presence of magic, at the risk of overstressing the nature and role of magical thinking and action in this context.

Using the ethnographic method of direct, on-site, and regular observations in the context of a laboratory, it is important to bear in mind the specificity of features of this fieldwork: the research has been conducted in France, in a secularized society and empirically speaking, in a scientific and engineering [End Page 486] environment that is not supposed to be conversant with expressions of faith or beliefs of any kind (except rationalist expectations concerning the machines' behavior). There are some precedents in the ethnography of laboratories: Latour & Woolgar were among the first sociologists to carry out empirical research in scientific laboratories and to describe the ordinary work of scientists and science "in the making," far from the abstracted model of scientific production.⁴⁰ Subsequently, Latour called attention on the process of "purifying" knowledge from "irrational" concretions, explaining why modernity purportedly excludes beliefs from the sphere of influence of science.⁴¹

Not all ethnographers investigating labs have however worked according to Latour's ideas, but contemporary ethnographic investigations in similar contexts now flow from two lines of development: "in the lab," the emphasis is on the construction of technologies and knowledge,⁴² and, outside the lab or "in the wild" (to refer to the technologists' lexicon), it is rather put on "users" and "usability" in ordinary contexts.⁴³ Whatever the case (labs in physics, biology, or engineering), the method of ethnography remains relevant in such contexts, when it's about the social and cultural factors shaping the products of "Science."⁴⁴ Robots, however, require a specific theoretical and methodological approach since they engage a reflection on the shape and the nature of the "human"; accordingly, anthropologists (such as Roberston or White in Japan, Becker in France, and Grimaud in India—see notes) have arranged their methodology so as to study the interactions between humans and machines in an ethnographic and non-experimental perspective.

In the same vein, I present here the results of more than one year of fieldwork in a laboratory of robotics in France (2021–2022), namely the Institute for the Study of Intelligent Systems and Robotics (hereafter ISIR), a public laboratory where different aspects of robotics are developed (exoskeletons, [End Page 487] mobile robots, robotic assistance for medicine, biomimetic robots).⁴⁵ For many anthropologists, this robotic model is potentially much more likely to generate emotions and beliefs more than any other machine.⁴⁶ Having had the opportunity to visit laboratories where both non-humanoid and humanoid robots are built and studied by roboticists, I noticed that understanding robots through the prism of "magic" seemed quite relevant in this context, given the resonance of the term in the literature and in public debates among roboticists.

When I first arrived in the lab, in September 2021, I introduced myself as "an anthropologist interested in magical thinking." Some of the members of the laboratory had already worked with anthropologists and were acquainted with certain preferred themes in anthropology (for instance, the role of culture in the history of technologies or of the body) and, above all, with the methodology of ethnography based on close observation of human behavior, contextualization, and critical analysis. I was warmly welcomed and presented at a large meeting of the laboratory, where the lines of my work provoked two reactions: either enthusiasm or caution, depending on whether the people I met saw the research as a kind of futile pursuit of the irrational ("magic? There is no such thing in here!" ironized many of them) or whether others felt that, on the contrary, it was time to reveal this less visible facet of the world of robotics ("if you are interested in magic, this is exactly the place" confessed the others).

I used a research methodology based on observations of everyday life, observation of roboticists' experiments with robots, interviews, photographs and films, and active participation as a full member of the laboratory (in my first year of observation) and as an associate visitor (in follow-up visits). During the time of extensive fieldwork, I collected empirical material consistently enough to provide hypothetically relevant elements for the operationalization of the notion of magic in the context of robotics, from an anthropological point of view. A regular presence in the laboratory was assured from September 2021 to September 2022, and was followed up by occasional visits. The ethnographic perspective consisted in recording everything the fieldwork researcher considered as relevant for the study, i.e. direct or oblique reference to magic. [End Page 488]

From the material collected, ethnography led me to consider three levels of manifestation of what can be described as magic in the anthropological sense of the term, i.e. as thinking and explaining events on the basis of causalities that have their own rationality, independent of the "objective" laws of the physical world. These three levels are: language, aesthetics, and practices. I will expose in more detail what kinds of ethnographic evidence fall under these categories in the following pages.

magic in language

Regardless of the specificity of the context (a modern lab) there is nothing surprising, in the anthropology of traditional societies, to state that occurrences of magic surface first in language and discourse, in the form of words, phrases, references and expressions; after all, classical anthropological studies consider magic as first and foremost an issue of words sustained by ritualistic performances.⁴⁷ Predictably, most of the recorded references to magic among roboticists at ISIR were discursive as well. It all started by spontaneous references expressed during face-to-face interviews or description of experiences by roboticists, related to the ethnographer-observer. When he first described his experiences in assistive robotics (i.e. body parts or organs being replaced by mechanical units), the director of the laboratory spontaneously stated that "this is what we are looking for, this effect [i.e. efficacy], this [the robotic unit associated with a human body] is an almost magical object since it is in command and adapted to the person's needs." Later, he alluded to a robotic exoskeleton as "working by magic!". The reference has regularly come back to me when roboticists were talking about the efficacy of machines—and some others were more critical ("forget about magic—this is all about technology!") or kept silent with a smile—a tacit expression of disregard. The ethnographer's interest did not lie in deciding whether this reference corresponded to the anthropological concept of magic: it was rather focused on magic as "indigenous category," exploring the different meanings it assumed for the individuals in the context of robotics.

As in traditional societies, magic here is a "floating signifier" in Lévi-Strauss's sense: a term relating to non-natural causes or agents.⁴⁸ It provides [End Page 489] a relevant repertoire in the context of extension and limits of the meanings and efficacy of technologies. Empirically speaking, "magic" surfaces in language through words, but not necessarily with a direct reference to the lexicon of the "supernatural" or the "irrational." A representational frontier between magic/ancient societies and rationality/modern society, however, exists in the mind of the engineers. In my first interview with a renowned French roboticist with an illustrious career, he depicted the men who first articulated and animated machines in history as "magicians of the past," whom he considered to be "pre-scientists": "they had no real knowledge of the subject-matter [of mechanics]." This opinion was almost a commonplace: dozens of interviewed roboticists adopted a similar stance and voiced similar terms when they were questioned about the possible role of magic in the historical and even contemporary developments of robots. Yet, this positivist standpoint was only officially supported by roboticists met during this study. Indeed, during the field study, the same persons were also inclined to voice anecdotes about those moments in their work when "strange" and even "thrilling" (in their own words) events have happened in presence of machines. Robots have often not behaved in the way their initial programming supposed they should have. To respond to the challenge allocating a sense to unanticipated behaviors, roboticists personify the attitudes of the machines by attributing a kind of intentionality to them—but intermittently. The above-mentioned roboticist discussed the case of a robot on wheels that was "running away" from the room in which it was tested, and it was convinced to "commit suicide" by running down a staircase. This puzzling attitude was solved when the roboticist discovered that symbols that the robot had to recognize to determine its trajectory with electronic sensors had, coincidentally, been placed … right next to the exit. "There's nothing magical about it," says the roboticist, smiling at me when he recalls the story (October 2021).

In other cases, however, things are not so simple and rational. In Japan, a year later, a young Canadian roboticist reported:

As a roboticist and scientist, I do not believe in these 'strange' things most of the time, but one evening I was alone in my laboratory, it was dark and quiet, and the robot I was programming had some unexpected reactions, started moving on its own. I knew it could have been a programming problem. But after a few minutes I was so frightened that I left the lab immediately and did not come back for three days.

(Interview with young male Canadian roboticist in October 2022.)

Even if roboticists adopt an official stance of rationality, there is room for alternative modes of thinking and experiencing that make the world "strange" [End Page 490]

Click for larger view
View full resolution

Figure 1.

articipative visualization table of associations concerning AI and robots. ISIR, Paris, 2021. Photo by Lionel Obadia.

and whose properties are close to those of magic; i.e. effectiveness attributed to non-physical causes, and in the last case, a kind of "ghostly" presence in the robot. This first mode of discursive expression is thus somewhat fleeting, based on a back-and-forth between the rationalist and "irrational" repertoires, alternating between explicit reference and implication, between disqualification and full recognition of these magical causalities. To some extent, then, magic is mainly at work here in an allusive way, as Davis, Fischer, Rose, or L. R. Bailey have also described.⁴⁹

The other aspect of language in which reference to magic has surfaced and captured the interest of the ethnographer is that of writing. A direct reference to magic was exhibited at the entrance to the laboratory until late 2023, materializing associations of ideas about robotics, intended for the audience who visited the laboratory in 2017 on an Open House day (see Figure 1). Consisting of a large whiteboard and keywords combined to form a sentence (subject—verb—complement), it was designed to weave threads [End Page 491] (physical and semantic) between different words about technologies and was proposed to visitors, non-specialists in robotics. Once the threads have been woven, the associations of ideas between the keywords were supposed to epitomize "profane" representations of technologies. According to one roboticist who conceived and assembled the display, interviewed in 2022, the aim of the public exhibition was precisely to "amend" some of these views by displaying "the truth" about technology "in the making."

One of the key terms, "artificial intelligence," had been associated to four adjectives: "magical," "mathematical," "futuristic," "scary," and "useless." The terms did not obviously refer to the same representations (some refer to the nature of AI, others to value judgements about it). Given the small number of threads running through the term 'magic,' it did not seem to have inspired visitors—this local empirical evidence was actually inconsistent with the global resonance of magic media. At first glance, one might think that the term was not a significant catchword at that time. But this option sounded too simple and did not do justice to the complexity of anthropological analysis, which requires a diversion through lexical categories. It was otherwise possible to assume that the complexity of "magic" might have hindered the visitors' choice of answer. I therefore asked the designer of the table what meaning she attached to this concept and what motivated her choice.

"To answer your question about the inclusion of 'magic' in the list of options in one of the questions, we made this choice inspired by Arthur C. Clarke's third law: 'Any sufficiently advanced technology is indistinguishable from magic'. As robotics is one of the advanced technologies, we were curious to see if magic was the visitors' favorite choice among the options included" (interview online, conducted in January 2022).

A third set of discursive occurrences can be found more directly in the lexicon of robotics and of the studies in Human-Machine Interactions (hereafter HMI). There are moments when roboticists invoke magic in derived yet explicit terms in an operational way, as part of their experimentation and testing activity: they allude to another dimension of the semantic field of magic, that of illusion, adding to the traditional features of wonder and instrumental efficiency. The most famous instance is the "Wizard of Oz," an experimental setup that is a widespread method in the study of human-computer or human-machine interactions.⁵⁰ The experiment is run by humans to make [End Page 492]

Click for larger view
View full resolution

Figure 2.

Detail of the table shown in Figure 1. ISIR, Paris, 2021. Photo by Lionel Obadia.

other humans believe that a machine is autonomous, intelligent, and responding to human behavior on its own. "Behind the walls" however, they are just humans manipulating machines, and by way of consequence manipulating humans. The Wizard of Oz is a method designed to test the mental processes of humans interacting with machines, in the most rationalist field of study: cognitive science. The reference is explicitly inspired by L. Frank Baum's children's fantasy, published in 1900, that became a movie in 1939. The main character, the Wizard, was actually an ordinary, powerless man who used tricks to make other people believe that he is a powerful and supernatural being. In the experimental setting, the reference to magic is not as trivial as [End Page 493] its namesake's fictional origin might suggest. It adds a level of stage magic to the previously exposed traditional magic, and roboticists allude to a similar intention and effort to distort reality so as to "make believe." This kind of deviation from science to fiction and fantasy is consequently both an allusion and an illusion unveiling the complexity of representation of magic among technologists—but still in the name of science.⁵¹

the aesthetics of magic

Art is undoubtedly the field in which the relationships between magic and technology are most salient: first, because art is commonly associated with emotions contributing to transcending the human condition through catharsis, and second because art echoes critical views on new technologies, especially those (like AI and robotics) whose promotion relies on the branding of their "magical wonder" or refers to the "magical" aesthetic of technologies, whereas the same technologies are also associated with dystopic scenarios and ambiguous feelings.⁵² In this regard, 'magic' surfaced on the emotional side of the relationship between humans and technologies, rather than the cognitive side only; therefore it is not exclusively linked to issues regarding knowledge anymore. Indeed, human desire to build technologies, has historically derived as much from a sense of wonder as from utilitarian concerns: from ancient times through the Middle Ages,⁵³ and into late modernity, both art and belief seemed to have played a crucial role in the relationship between humans and technologies. While robotics laboratories are not places where pieces of art (in the strict sense of the term) have been created or exhibited, they are explored by artists for aesthetic projects like photographs or movies of the machines; artists also collaborate with roboticists to promote their productions through photo exhibitions (arts that include robots), or robot exhibitions (robots as the main subject matter).⁵⁴

The aesthetic of robots has infused many media since the beginning of the twentieth century,⁵⁵ and occasionally, robots in laboratories have been [End Page 494] disposed into artistic performance.⁵⁶ Guided by an anthropological training, the ethnographer could also consider the laboratory as a cultural space replete with symbolism, with special reference to the material arrangements and the place's aesthetics. Reaching this laboratory for the first time, as an ethnographer with a methodological tendency to focus on details, I was struck by the chaotic material organization of the workplace of those who design, create, and test robots. This is especially true for material fragments and wires, tools, and batteries, but also for entirely complete robotic prototypes: few of them were plugged in, and some were simply discarded in the corridors of the lab. All of the rooms were filled with pieces of machines, arms, hands, metal rods, joints, computers, buttons, cameras, etc., in various states of assembly and testing, and a multitude of tools and boxes were lying all around. The work of roboticists is to build, assemble, and associate, to a certain extent, in the way of craftsmen, ancient apothecaries, and even alchemists—the metaphor is a fashionable reference today. This association of ideas came to me after an observation of practice in medical robotics: I was watching a young researcher concentrated on a task, alone in the basement of the laboratory's experimental area. The light was low, and he was concentrated on completing a difficult task (programming part of the sequence of a robot's movement); he did not pay attention to my presence or to any other element in his environment. Almost immobile in front of the machines he was programming, he was surrounded by the many objects needed for manipulation (see Figure 4).

By analogy, the ethnographer's mind quickly made a connection with medieval and renaissance paintings of magicians, and in particular The Alchemist by Cornelius Bega (1663) (see Figure 5): the composition of the scenography, the position of the bodies, the distribution of the objects in space, the saturation of the frame by objects of different size, kind, nature, and form, inspired this comparison. In both cases, one can see a solitary individual, concentrated on a task, surrounded by small and large material items, and both share the objective of producing a material with effective properties that has the power to transform the world. Graphically, the correspondences are striking, regardless of the historical and cultural contexts that distinguish the two images. Comparison makes robotics laboratories resemble old-fashioned "cabinets of curiosities" filled by all shapes and sizes [End Page 495]

Click for larger view
View full resolution

Figure 3.

A young French roboticist in action. ISIR, Paris, France, 2021. Photo by Lionel Obadia.

of objects. This impressionistic experience paved the way for a deeper reflection on the technical nature of magic: should the scope of magic be limited to certain techniques that are required for performative efficient actions (ritual techniques and materials utilized in magical ritualism), as Mauss suggests,⁵⁷ or can it be extended to any kind of human technology?⁵⁸ By extension, are the formal similarities between the aesthetics of alchemy in practice and robotics in action sufficient to ensure that they are of the same nature? Of course, there is a risk of methodological gaps and conceptual dead-ends in these comparisons between ancient paintings depicting magical practices and contemporary experiments in the materiality of robotics laboratories. They indeed result in an interplay of mirror-images that postulates, before even [End Page 496]

Click for larger view
View full resolution

Figure 4.

Cornelius Bega, The alchemist. Dutch Republic, 1663. Image courtesy of Alamy.

being demonstrated, that the observable similarities between human attitudes and material arrangements would end up in concluding that they are different facets of an identical reality. The particularities of contexts and material/symbolic forms in which they take place are overlooked. Moreover, in the laboratory, magic is mainly allusive and only surfaces sporadically, whereas it is at the very heart of the alchemist's daily occupations. Whoever wants to establish on solid grounds the empirical existence of forms of magic, needs [End Page 497] to bring in material other than aesthetic images, and therefore to turn to practices, since magic traditionally unfolds through gestures and rituals.

magic in practices

This appeal to the most classical approach in anthropology, magic in and as practice, was an attempt to grasp its empirical occurrences and to link them, as far as possible, to mental representations expressed through discourse (what scholars used to refer to under the generic term 'belief'). In this laboratory, as in the others I visited in the university of Waseda, Japan, reference to magic could also be deduced from behaviors and attitudes when roboticists interacted with the material and the mechanical elements that make up the robots. But this classification of the attitudes falling under the category of magic is challenging. On the one hand, indeed, roboticists professed rationalism, such that any form of belief, including 'magical' ones, were excluded from the processes of designing, creating, developing, testing, and experimenting with robotic machines. These tasks were supposed to involve only formal knowledge—mathematics, mechanics, electronics, and, of course, computing. Magic was in this case a by default category—it explained only residual elements that could not be elucidated by rationalist repertoires. On the other hand, reference to magic sometimes reflected roboticists' openness to the idea that something 'irrational' could exist and surface in robotics, insofar as not everything is rational in human attitudes toward technologies, and leaves room to "marvel" and for "irrationalism," by excess.

During fieldwork, reference to magic filled the gap of unexplainability of technologies, and offered a possible but not systematic interpretation, that roboticists amusingly attributed to each other through tactfully voiced anecdotes:

I do not believe myself in these things, but other young scholars in the lab do, but they will not tell you anything—they do not want to look like irrational minds. They play heroic fantasy games at night and adopt strange behaviors in private when they interact with small robots—they caress them and talk to them!

(Interview with young female scholar at ISIR, 2021).

As if magical thinking could only exist through the eyes of one's own rationality mirrored by that of others—as was the case of folk magic of traditional societies transfigured through the prism of modern and Western rationalism.⁵⁹ If ever there exists a "magic of robots" it has something to do [End Page 498] with the enthusiasm expressed by humans when they see the machine in action: "it works by magic!" Interestingly, the logic of "magical" association was particularly salient in the case of assistive and prosthetic robotics. Another example can highlight this point: Christophe, a mature man who was born with only one upper right arm, is the official "tester" of arm prostheses in the laboratory: on a regular basis, he is coupled with mechanical arm extensions using a sophisticated system of sensors for electrical impulses produced by his muscles (see Figure 5). From the beginning, he has always maintained that magical thinking was "very important" to him in his relationship with the technologies that helped him to augment his body's abilities and grasping capacities. In his view, "magic" refers to the mental operations that allow him to hybridize with mechanical elements and accordingly, to embody the "magical" synthesis between human and machine.

Click for larger view
View full resolution

Figure 5.

Christophe H., testing a robotic arm. ISIR, Paris. Image courtesy of ISIR.

[End Page 499]

Another irruption of magic into robotics consists of moments when machines fail to meet human expectations, and when the frames of experienced reality break down. These technological or computational ruptures—be they called "glitches," "bugs," or "errors," are of course well known in the domain of computing and engineering robots. Programmed, prepared, tested many times, the robots are technically equipped for a precise action or reaction on command, but they do not always respond or respond correctly to the computer requests and human expectations. This is what roboticists in laboratories call the "l'effet demo" [the demo effect], which describes the unforeseen and disturbing technological problems arising in the context of a technological test in the presence of an audience. These problems may arise when a robotic device is programmed and staged at a "demo," an exhibition devoted to testing the technological performance of a robotic prototype in front of an audience of primary importance—other roboticists, representatives from the company that has commissioned funds for the robot, prospective clients, research funders, scientific authorities, and journalists. As seen through anthropological goggles, a demo is a highly ritualized and therefore institutionalized moment for which the robot has been trained hundreds and even thousands of times. Alas, it often happens that it is precisely in these moments that the machine malfunctions. All the roboticists I've met in this laboratory and in other places or countries have at least one (but often many more than one) account about robots remaining motionless when they were expected to move, or moving in a different direction from the one intended by means of computation, and finally that can even hurt or harm people when they are deployed and in contact with humans. In ISIR, in October 2021, I observed a demo of the remote-controlled articulated arm Jako©, designed to perform simple tasks for disabled people, but the robot began to strangely repel human bodies as if it was "angrily" trying to move them away from its trajectory and was therefore gifted with human consciousness and feelings. A couple of months later, a similar "demo effect" was recorded during a test of a Baxter humanoid robot, which is used for scientific purposes: programmed to reproduce human gestures, the robot has been trained thousands of times to throw a tennis ball into a bucket. But when I stood in front of it and the engineers and researchers tried to activate it to show me what it could do, the robot suddenly began to systematically fail. With a smile of complicity, one of the engineers turned to me and said: "well, you see, it's really magic. I bet your presence confuses it." They then set out to solve the problem by checking the computer program. Magic was, in this discursive context, allusive and witty. [End Page 500]

Click for larger view
View full resolution

Figure 6.

Baxter brand humanoid robot at the ISIR laboratory, in the context of a demo. ISIR, Paris, 2021. Photo by Lionel Obadia.

The experience of a technological bug nevertheless unfolds different repertoires for interpretation, but in this case, non-technical and non-rational causes were not deemed relevant and applicable: they are considered as trivial and supported only by non-specialist ordinary "layman" audiences, while roboticists favor (officially) only rationalist explanations like computer bugs, programming issues, connection problems between the program and the robot, and so on. In private interviews, however, several engineers conceded that they had been tempted by the possibility of introducing non-rationalist elements in their experiences: allusions can equally refer to "evil spirits," "supernatural forces," or "ghostly presence," recalling the properties of mana, i.e. magical thinking in the context of "primitive societies," associating ideas in a continuous process of assigning meanings without fixing any of them.⁶⁰ Interpretation voiced by engineers followed more or less similar patterns, not only because of, but also despite the lack of stable explanation for the robot's [End Page 501] malfunctions. If a "magical explanation" (i.e., an explanation referring to no physical and logical causes) could find a way in the rationalist world of a robotic laboratory, in this context it remains all but the main reservoir for meanings, a function it assumed in ancient non-Western societies. It however remains relevant as a potentiality, allusive and optional, once the other options to elucidate unexplainable robots' acts and movements have been exhausted.

in conclusion: "empirical testing," and the virtues and limits of analogy

What can we draw from these ethnographic elements to reflect on the magic at the heart of the most sophisticated high-tech production spaces in modern societies? Anyone who imagines robotics laboratories are spaces crowded by "intelligent" and "autonomous" machines interacting rationally with humans in settings inspired by science fiction movies is obviously mistaken. Of course, a robotics laboratory is first and foremost a place described by those who work in it as governed by the ethics of science, where machines are developed according to the principles of mathematics, electronics, and mechanics.⁶¹ As a social entity, it is replete with beliefs, values, and worldviews of those who work in these particular places. Among the community of roboticists, some shared ideas are inspired by the cultural norms of scientism prevailing in robotics, others are culturally inherited, and others pertain to psychological perceptions and reactions in the context of interaction between humans and machines. Many of these ideas and attitudes relate, in one way or another, to "magic." Contemporary media fascination with the 'magic of technology,' as well as the trendy fashion of high-tech companies that are named after the word 'magic' or even use references to magic as a merchandizing strategy for their products and devices, suggest that robotic technologies partake in a re-enchantment of the hypermodern world. Ethnographic evidence calls for a more cautious conclusion.

In the context of this qualitative and localized study, complexity must indeed prevail over an over-simplistic stance about the "magic of robots." The several empirical elements extracted from the field study establish that if "magic" comes into sight, it is expressed through many different types of discrete signs (linguistic, practical, symbolic, or aesthetic), all alluding to alternative and non-rationalist ways of explaining events. Rather than jumping in the contemporary debate about the relevance of magic in today's heavily digitized world, there is an obvious need to consider first and foremost to which conditions this term is applicable (or not) to seize and give sense to [End Page 502] the fleeting outbursts of the unexpected and unexplained aspects of reality in robotics.

Be it allusive or metaphorical, as it appears in the media, magic is far from the first level of signification to surface in a roboticist's laboratory life. A closer and detailed look calls attention to the multitude of small events and discrete signs unveiling the presence of magic, risking, of course, a partial and biased approach to the empirical data as seen with the goggles of anthropology of religion. Reference to magic in technology and labs is not utterly new, but it still is not documented enough. Stef Aupers paved the way with his remarkable study of Silicon Valley,⁶² and in another vein, Maciej Musiał has done so in robotics.⁶³ Such perspectives open up fecund avenues of research: the material collected indeed unveils formal similarities with the magical thinking of traditional societies, and reflects on possible functional and symbolic similarities between them. "Magic" thus needs to pass the test of empirical relevance before being otherwise discussed at length. Yet, it also calls for an empirically based and contextualized definition of "magic" since it is not a ritualistic magic that is being expressed here, but rather a kind of magical thinking that is resistant to the assaults of Rationalism in modernity and seems to find its way in into the scientific milieu.⁶⁴ [End Page 503]