Anthropology's Sad AI Archive

 

There are 3 approaches to generative AI in the classroom: 1) an outright ban on it; 2) a limited use policy that covers certain assignments or parts of assignments, and 3) an open approach that allows students to do what they would. None of these are fool-proof, whatever the intentions of the professor. Ultimately, generative AI are third-party, black-boxed products–more tempting to students, perhaps, than Wikipedia, but also more treacherous. I feel for my colleagues in the humanities attempting to wrest essays from students on Shakespeare or Aristotle: generative AI is all too good at producing a mediocre essay on these subjects. I also understand my colleagues in the computer and information sciences, who utilize these chatbots to help with their instruction.

 

But with anthropology, there are several caveats. 99.99% of writings on other peoples of the world are drenched in ethnocentrism, colonialism and racism. The internet is awash in complete nonsense about “tribes'' and their “traditional culture,” and, in generative AI, all of this is ground up and, like sausage, pumped into prompt-driven content. Yet typically, students don’t know enough to be able to distinguish a “good” and “bad” response from ChatGPT or Gemini. 

 

This is a somewhat longer way of saying that students often tried to utilize generative AI in my introductory assignments and take-home exams, and their grades suffered for it. Not because I was penalizing them for cheating; proving that they’ve used AI is almost impossible, and generative AI detectors are unreliable at best. Instead, the questions that I asked were all about the anthropology I’ve taught in classes, and generative AI is, unfortunately, only too willing to spit out all manner of palaver. Only someone who knows what to ask can minimize the racism and colonialism inherent in generative AI engines. The default is ideology. And hallucinations. 

 

One thing I want to include next year is some process of education. I really think that students don’t really know any better. The least I can do is show them that it’s not so easy and explain why that is–that generative AI is not giving them the “truth.” Or, rather, it is: the truth of colonialism and racism that underlies Western thinking about non-Western peoples. Anthropology’s sad archive. But to someone who’s never taken anthropology before, this stuff looks correct to them, and the temptation is too strong, especially in the panicked moments before a deadline.

Multimodal Methods in Anthropology

Today (April 26, 2024), our book, "Multimodal Methods in Anthropology" is released into the world. Here's a song I've created for the moment using Udio, a text-to-song Generative AI model: https://www.udio.com/songs/m5HMHSZ2exSgEWE7f8AaAr

And here's a code for a discount on this book from our publisher, Routledge Books:

Anthropology's Seen and Unseen

In 1866, Alfred Russel Wallace sent a breathless letter to his friend and colleague, Thomas Huxley, inviting Huxley to join him in exploring a “new branch of anthropology”: Spiritualism. Based on the explosive growth in seances in Europe and North America, Spiritualism conjured a world beyond the grave where the dead continued to learn and improve themselves while enjoying each’s ghostly society. When the dead deigned to visit the living, it was typically to help and to advise: “this haunting could teach the living how to build a more perfect society in the here-and-now” (Forbes 2016: 445). So, for Wallace, this really was a new direction for the field–an inquiry into the unseen and, simultaneously, into a future that awaited humanity, both in the afterlife and in the perfection of human life on this planet.

 

To give away the ending, Wallace was not successful in establishing his “new branch of anthropology.” Huxley wasn’t buying it, and other anthropological colleagues eventually turned against him as well. After wavering a bit through his own field investigations of London seances, E.B. Tylor weighed in against Wallace, relegating Spiritualism to a “survival” from more “primitive” times. But Spiritualism held some fascination for Tylor as well, even as he tried to distance anthropology from it. In the end, though, what would become “anthropology” as we know it would only engage with Spiritualism as a symptom of something else, in the same way that magic, religion, and ritual would yield to an understanding of deeper truths. Or is there something more? As Pels and others have argued, this episode reveals something more about the way anthropology regards its object.

 

On one hand, the two couldn’t seem more different. On the other, anthropology and Spiritualism–both middle-class endeavors arising in the context of 19th century empire, suggest a variety of homologies. The most salient, perhaps, is the idea of culture itself. Yes, “culture” surrounds us in countless material forms–but at the same time, it does not. A behavior, an object–to be sure, these are “cultural,” but where is that “culture”? As Delaplace points out, “Describing “culture” should also include an actual account of its “wholeness”: the invisible thread, as it were, which bundles up these cultural components into a totality” (2019: 37). In other words, “culture” is the results of the anthropologist’s revelation, the end results of an analytical process that renders an unseen world of connections and isomorphisms visible to the anthropologist’s audience. Spiritualism would make the same claims–the technologies of the seance were precisely those revealing a concealed world: spirit rapping, table levitation, automatic writing, spirit photography.

 

And, like Spiritualism, anthropology also involves concealment. During the 19th century, the Spiritualist medium utilized a number of contrivances–dimly lit rooms, screens, capacious tablecloths. Whether or not you believed in Spiritualism, these were the preconditions for the spiritual knowledge. For anthropology, the trick of culture was the revelation of relevant behavior and the concealing of what was considered extraneous. In photography, for example, “the epistemic virtue of ethnographic photography entailed the ability to hide certain things” (Delaplace 2019: 39). Franz Boas and George Hunt would work to (playfully) conceal the evidence of settler colonials from their portraits of indigenous life. And Bronislaw Malinowski and Margaret Mead would both conceal the impact of evangelical Christianity on people at their fieldsites, all with the ultimate goal of representing “authenticity” in native life.

 

This kind of concealment, of course, is no longer part of anthropology. Or is it? When we look back at the prevalence of reflexivity in late twentieth century anthropology, we might be led to believe that anthropology had gone all-in on revelation. Anthropologists located themselves amidst intersectionality and intersubjectivity, lifting the curtain and turning on the lights, as it were. There were still concealments, however. Or, to be more exact, the revelations of reflexivity facilitated other dimensions of the unseen. Consider, for example, the dominance of a handful of elite departments in producing the vast majority of academic anthropologists, and the embeddedness of those universities in structures of US empire (Speakman et al 2018). Or, alternately, the similarities between anthropological work and extractivist industries commodifying indigenous knowledge and practices for a rapacious capitalism transforming all life into exchange value. These undercurrents are concealed, as the precondition, perhaps, for the revelation of other anthropological “truths”.

 

There is, therefore, a dynamic in anthropology that we can trace from the 19th century–one that extends between the seen, the unseen, revelation, concealment and, as Taussig had written, “the skilled revelation of skilled concealment” (Taussig 2003: 273). To be sure, this configuration changes over the course of anthropology's history. But it’s the imbrication of anthropology in this dynamic that betrays anthropology’s embeddedness in capitalism and western imperialism as a whole for, as an economic system, capitalism depends on concealment for its strength: the alienation of labor, the destruction of the global south and, ultimately, the untimely end of human life on this planet. All of these must be concealed for the revelation of value and the novelty of production.

 

Marx, of course, realized this at the fore, and his Capital contains references to the very same Spiritualist practices that perturbed Wallace and Tylor in the 1870s. A table is “just” some wood joined together, he explains:

But as soon as it emerges as a commodity, it changes into a thing which transcends sensuousness. It not only stands with its feet on the ground, but, in relation to all other commodities, it stands on its head, and evolves out of its wooden brain grotesque ideas, far more wonderful than if it were to begin dancing on its own free will. (Marx 1990: 163-164)

This was, of course, an allusion to the “table lifting” practices in seances. The commodity itself issues from the dialectic of revelation and concealment, and the 19th century’s human sciences soon craft their own “dancing tables” discerning culture, society and the psyche through the very same transformational calculus. Yet even in this critical revelation, we would be wise to think about the concealments that enable this insight.

Communication without Control: Anthropology and Alternative Models of Information at the Josiah Macy, Jr. Conferences in Cybernetics

 

Communication without Control: Anthropology and Alternative Models of Information at the Josiah Macy, Jr. Conferences in Cybernetics

 

Samuel Gerald Collins

Towson University, USA

scollins@towson.edu

 

 

The characteristics of our digital world—algorithms, virtual reality, AI, cryptocurrency, etc.—were largely formulated during the Josiah Macy, Jr. Conferences on Cybernetics, held between 1946 and 1953. The concept of reducing the world to flows of information is one of the legacies of these meetings, with all of the alienation and ideological work that “the digital” has perpetrated. Yet there were anthropologists at the Macy Conferences as well; Margaret Mead and Gregory Bateson attended every meeting, and recent scholarship (e.g., Geoghegan 2023) has shown how anthropological thought contributed to the formation of our digital world through the reduction of culture and social life to codes and feedback loops. Yet there were also alternative models proposed during the Macy conferences, e.g., an embodied model of information championed by Donald MacKay (Hayles 1999). This paper looks to another alternative, one based in misunderstandings at the Macy conferences themselves. In practice and in discourse, Mead and Bateson held very different ideas about what “information” could mean—ideas diverging from the “command and control” model that would predominate. Despite those differences, though, the Macy conferees could communicate with one another and even plan projects together. This is their model of information—communication without reduction and without perfect understanding—interfacing rather than dominating. Anticipating the work of British cyberneticist Gordon Pask, this model of anthropological cybernetics opens the possibility of a communicative informatics without control, where interaction can develop without reduction and understanding without domination.

 

The Josiah Macy, Jr. Conferences on Cybernetics (1946-1953) were borne on the a crest of Unity of Science discourse in the United States, and sought to bring together the social sciences with the physical and material sciences through the development of a “new lingua franca” where, as Kline writes, a “universal language of information, feedback, and homeostasis” could “model all organisms from the level of the cell to that of society” (Kline 2020: 13). The meetings were remarkable for many reasons, not the least of which was the active participation of the anthropologists Margaret Mead and Gregory Bateson. It was one moment where anthropologists would speak as equals to their hard science counterparts in a nationally prominent forum. That was the promise of cybernetics for anthropology–a place at the table to shape the direction of scientific and technological development in the age of American empire.

However, in all these endeavors, cybernetics failed. After percolating through multiple disciplines and through popular culture, cybernetics–perhaps, as Geoffrey Bowker has termed it, through a process of “legitimacy exchange”--lost currency as scientific discourse (Bowker 1993). By the 1980s, very little in the sciences and social sciences was being produced under the auspices of cybernetics, and most new research (along with this paper) represents historical analysis rather than new applications of, as Wiener called it, the “the science of control and communications in the animal and machine” (Wiener 1948). In anthropology, there was very little mention of cybernetic theory in anthropological work after the 1970s, even as anthropologists began to turn to considering the digital life of the cyborg.

And yet, there’s also ample evidence that cybernetics succeeded–too well, as it turns out. All of the ideas that catalyzed the Macy Conferences: game theory (von Neumann), neural nets feedback and circular causality (Wiener, Rosenbluth and Bigelow) and information as a ratio of signal to noise (Shannon)–have led to the triumph of the virtual and reduction of the world to digital flows of variously commodified information. In no small part, we can credit the Macy Conferences for a world where teaching is reduced to the digital delivery of content, a world where AI agents proliferate in our lives, threaten our careers and prompt our constant  “Turing suspicions” that the people with whom we digitally interact may be artificial agents instead. As Bateson suggested later, “cybernetics” has taken on the association of “control” at the expense of communication (Bateson 1991). It has become the blueprint of our digital domination; whoever controls the information controls the world.

And it was successful in another, more covert sense. Geoghegan’s recent book, Code, eloquently follows the break-up the Macy Conferences into its constituent disciplines, with the triumph of information leading anthropologists down a road to semiotics and structuralism, while circular feedback underlies the continued popularity of Cognitive behavioral therapy (Geoghegan 2022). As a potent, generative metaphor, “code” has structured the ways we relate to the world and each other, but in ways that have ironically widened the gulf between disciplines.

There were, of course, alternatives expressed at the Macy Conferences. As many theorists (including Gregory Bateson, Niklas Luhmann, Humberto Maturana, Franciso Varela and others) have noted, second order cybernetics grows out of the unanswered questions about the role of the observer in the system, while Hayles (1999) looks to Donald MacKay’s more embodied understanding of information as an anodyne to the digital’s hostility to the physical world of nature and bodies.

Yet, despite Mead’s and Bateson’s prominence in the Macy conferences and to cybernetics in general, there has been little explicitly anthropological about the ways our digital lives have developed, even as anthropology has turned to the study of those lives as objects of our research. But it’s not for lack of trying. The Macy Conferences transcripts (however spotty and incomplete) show Mead and Bateson engaging in the work of the anthropological gadfly–tempering the strident parochialism of the Macy attendees with exceptions drawn from the anthropological record. As Mead sagely notes to Lawrence Kubie’s universalist proclamations on the role of the unconscious, “If you look at some other cultures, you don’t necessarily find that same contrast” (Pias 2016: 426). But the exception rarely derails the Mcy conferees who are, after all, busily re-mapping the world in the context of the post-War US empire. So while Mead and Bateson contributed commentary, central ideas (neural nets, information as signal) escaped the Macy conferences as universals unhindered by ethnographic exception.

Nevertheless, there are unacknowledged anthropological contributions, and this paper concerns one of them: the role of misunderstanding. At first glance, this is not a very promising alternative–it seems antithetical to the whole point of the conferences, after all. The lingua franca of circular information, circular causality and neural nets was supposed to unify knowledge by describing universal processes as characteristic of chemical reactions as they were of human culture. And yet there were many misunderstandings. For example, Ralph Gerard was, reportedly, “intensely frustrated by the perpetual tangents to tangents that developed during a meeting and the rare satisfaction of intellectual closure and completion of any line of thought or argument” (quoted in Hayles 1999: 73). For their part, Mead and Bateson pilloried the obstinacy of their fellow conferees in a 1976 interview with Steward Brand: “”So we used the mode “feedback,” and Kurt Lewin–who didn’t understand any known human language, but always had to reduce them to concepts–he went with the idea of feedback as something that when you did anything with a group you went back and told them later what had happened [ . . . .] So the word ‘feedback’ got introduced incorrectly into the international UNESCO type conference where it’s been ever since” (Brand 1976: 5).

This paper is concerned with something I’m calling misunderstanding - not as the failure of cybernetic communication, but as its precondition. Anthropologists play the role here of not reducing things to her underlying code, but of facilitating interaction between different ideas without introducing a third term. Perhaps re-framing is a better word? After some examples from the Macy Conferences transcripts, the paper develops an alternative model of communication resembling Andrew Pickering’s emphasis on performative ontologies. I preface this by noting that these exchanges are, in many ways, what critics of the Macy Conferences found most irritating: the digressions and non sequiturs of meetings that went frequently off the rails yet were still preserved in transcription. Or, not all: Hayles notes that Teuber blocked publishing the transcriptions of the final meeting in 1953, “noting that the discussions were too rambling and unfocused; if published, he said, they would be an embarrassment” (Hayles 1999: 76). This does not mean, however, that the other transcripts were tightly focused.

Two examples:

In the 1949 Macy Conference, Lawrence Kubie gives his paper, “The Neurotic Potential and Human Adaptation.” Later discussion prompts Kubie to ascribe neurosis to all forms of symbolic communication: “The capacity to communicate by means of language symbols, and the capacity to become neurotic are very close together” (76). Mead, in her capacity as anthropological elder, shifts the conversation to culture and learning processes: “These expectations may be different from another culture which expects another area of behavior to be subject to learning: in any given society the particular acts, the particular behavior in which conscious and therefore flexible learned behavior is to play a given role may vary” (76). Kubie seems unsure of Mead’s point, and reiterates his initial claims, but this seems to only draw out discussion to the possibility that autonomic responses could constitute an unconscious, or that societal expectations for communication constitute what we regard as neurotic. Eventually, a frustrated Kubie continues trying to wrest the conversation back from culture and statistical modeling to the primacy of the psychological. “I am puzzled here because I am not quite sure how this confusion has arisen. Let me retrace my steps for a moment” (80). A bit later, Kubie’s colleague, Henry Brosin, gives him “a chance to catch his breath”--and we can imagine, perhaps, an apoplectic moment for the orthodox Freudian (82).

Another example, following on J.C.R. Licklider’s 1950 paper, “The Manner in Which and the Extent to Which Speech and Can Be Distorted and Remain Intelligible.” Licklider’s contribution is really a paper on sound engineering–with multiple graphs illustrating sound clipping and frequency. Wiener brings up the problems of translation between human and machine: “I  am considering, for example, the remote-control substations for hydraulic power, where the power dispatcher has to get messages to the machine and where the machine has to inform the power dispatcher about significant facts. Now there again you have the same translation problem. (227)

Mead, however, picks up the question in terms of human communication, and other conferees follow with their own anecdotes. Can people communicate across languages in a related family and across dialects? Her question: “I should like to get back to the question: Is this translation or isn’t it? What is translation?” (237).  It is, Mead concludes, “a question of framing. Some people frame all European languages together, so that the idea of learning any Indo-European language is regarded as just another language to be learned; the idea of learning Hawaiian or Japanese, however, would be regarded as something quite different. What is translation for one person is not translation for another” (237). Other examples follow, turning to physiology, to psychology, to hypnosis, speech pathology, etc. Finally, with a series of anecdotes about memorizing strings of characters through song, discussion ends–some distance from the sound engineering where it began.

Discussion

Anyone who has been to an academic conference will recognize these in these transcripts characteristic verbal interactions: questions that are really comments, critiques that are little more than self-aggrandizement. What makes them different here is their relationship to the goals of the Macy Conferences. Each of the concepts that fuels these cybernetic meetings–neural nets, feedback and information–is taken up again and again by conferees, where conversation spins them into unexpected and occasionally refractory directions. Attendees are talking, but they are not moving towards a consensus. This would be a searing critique of the Macy Conferences, if, indeed, the point was to create consensus. But if achieving consensual understanding was not the point, than what was?

A later example: the second-generation cyberneticist Gordon Pask built an installation called the “Colloquy of Mobiles” involving five robots (two “male” and three “female”) rotating towards each other on the bases of lights and sounds. What marks this off from other cybernetic machines, though, is its performative dimension. Pask meant for humans to interact with the Colloquy. As he wrote, “the mobiles produce a complex auditory and visual effect by dint of their interactions. They cannot, of course, interpret these sound and light patterns. But human beings can and it seems reasonable to suppose that they will also aim to achieve patterns which they deem pleasing by interacting with the system at a higher level of discourse” (Pask 1971, quoted in Pickering 2010: 359-360). Eventually, human visitors to the Colloquy began using small mirrors to reflect lights back to robots in order to evoke a response. Not “communicating” with the machines in the sense of a shared message but interacting with the Colloquy in order to produce light and sounds that the human visitors found pleasing.

Was the purpose of the Macy Conferences to reduce the world to processes of information and circular feedback? Or were there other more interactive and less dominating alternatives?

Another moment in the conferences, this one in the wake of Bateson’s 1952 paper, “The Position of Humor in Human Communication.” By now, random asides are no strangers to these proceedings, and many of the conferees take a moment to tell their own bad jokes. John Bowman then suggests that humor could be produced in a machine using a simple circuit: “A circuit of that type may have two stable states. If it is put in any state, it will asymptotically approach one of the two stable states and stay there. On the other hand, with the same components in slightly different values of the circuit constants, it can oscillate” (548). To which Bateson replies: “I am always prepared to say that an electronic buzzer is laughing” (548).

Does Bateson mean that both circuit and the human are laughing? Is he suggesting that laughter is nothing but the oscillation of circuits? Or that it is the confusion of signals resisting homeostasis? One of the legacies of the Macy Conferences has been precisely that: the reduction of physical to the information that lies beneath. This is also, as Geoghegan notes, the underlying mechanism in the reduction of social and cultural life to its semiotic exchange. There are, however, other possibilities. What if the “laughter” in the person and the “laughter” in the circuit weren’t equivalent, after all? And what if the oscillation in the circuit was not the oscillation in the neural stimulus? What if that was just something we were calling laughter and oscillation in order to create an interactive bridge between two dissimilar systems? This is certainly more likely than Bateson becoming a reductive materialist, and moreover anticipates the work of Gordon Pask and other second-order cyberneticists.

Figure 1

 

(Figure 1: Shows two systems interacting without a) dominating each other; or b) being subsumed under a third term)

This sense of cybernetics acknowledges a world that is not reducible to defined quanta, yet one in which we very much need to interact in order to live. And this is the world of the Macy conferences, where conferees, perhaps, have little to show for their efforts after 10 years of meetings and endless conversations. And yet they still had those 10 years and endless conversations, despite their intractable differences between their disciplinary homes. In what Mead later summarized–rather unhelpfully– as ”microevoltuion,” “The conversation would come to an end and be resumed. There would be freedom to talk and freedom to listen, and the web of meaning would be woven as we talked, making a new pattern before our eyes” (Mead 1964: 301).

In other words, this cybernetics as a means of interaction across difference–however defined. As Pickering writes, “the cybernetic sense of control was rather one of getting along with, coping with, even taking advantage and enjoying, a world that one cannot push around in that way” (Pickering 2010: 383).

At present, we are living in just such a world, one where countless nonhuman and multiagent systems seem to go their own way despite our efforts. Those agents may be variously intelligent. With generative AI, we have agents that simulate human conversation, so it is easy for people to (mis)understand their agency. Most algorithms, however, toil in the background, running checks on our job applications, checking purchases against credit history, touching up photographs. Yet in all cases, it is not so easy to understand how multiagent systems make decisions. Large language models, for example, are unable to tell us which sources they’ve consulted in their replies, and text-to-mage generation has notoriously surreal moments. We may describe those systems according to a number of anthropomorphic metaphors, and, similarly, ascribe any number of information processing functions to ourselves in return. But, to be clear, neither of these is an accurate representation.

This is the world we live in now, but it is not a “post-human” one in the sense of a reduction of human life to manipulable flows of data. Instead, we find ourselves enmeshed in interactive systems that can’t be reduced to information quanta–even if powerful organizations strive to do so. Amidst these heterogeneous systems, communication may be an impossibility; instead, coordination may be the optimal outcome. The basis of that coordination is the questions: who defines the field for interaction? Will everything be subtended to profit? State surveillance?

On the other hand, we might strive towards making these systems explicable and more predictable, more interactive, more responsive and more grounded in the experiences of diverse communities: “Explainable artificial intelligence.” “community robotics,” community-based design. Anthropologists are uniquely poised to consider these networks of humans and non-humans as they interact together to create their world. They are also uniquely positioned to insist that non-human agents need not only serve the whims of capital, and that rejecting information as “command and control” means opening up recombinatory possibilities.

More than this, what does it mean to interact with the multiagency around us as an equal partner? What does it mean to describe the give-and-take of a horizontality in cybernetics?

References

Bateson, Gregory (1991). A Sacred Unity. NY: HarperCollins.

Bowker, Geof (1993). “How to be Universal.” Social Studies of Science 23: 107-127.

Brand, Stewart (1976). “For God’s Sake, Margaret.” Co-Evolution Quarterly (Summer): 32-44.

Hayles, N. Katharine (1999). How We Became Post-Human. Chicago: University of Chicago Press.

Kilne, Ronald (2020). “How disunity matters to the history of cybernetics in the human sciences in the United States, 1940-80.” History of the Human Sciences 33(1): 12-35.

Pias, Claus, ed. (2016). Cybernetics: The Macy Conferences, 1946-1953. NY: diaphanes.

Pickering, Andrew (2010). The Cybernetic Brain. Chicago: University of Chicago Press.

Wiener, Norbert (1948). Cybernetics: Or Control and Communication in the Animal and the Machine. Cambridge, MA: The MIT Press.