A memory bank of the future: Stiegler, education and the gesture of care

Abstract

In contemporary societies, the processes of transindividuation by which knowledges are transformed into cycles and rhythms of metastability have been dramatically short-circuited. In turn, this has provoked the spiritual misery and pseudo-fabulations so prevalent all around us, including our educational contexts. For Stiegler, this is nothing short of a noetic reticulation that deprives us from ways of thinking ourselves beyond or outside of our digital experience. But digitality has not only intensified the commodification of knowledges (savoirs), it has also rendered even knowledge production automated, recursive and probabilistic, the uncritical implementation of ChatGPT being a prime example. What this means is that knowledge and knowledge production have been subsumed under the rubric of recursive optimization for predictive performance. To understand this transformation, I discuss the implications of the widespread use of Bayesian statistics in machine learning. My argument is that we need to develop new speculative tools aimed at what is known as priors in Bayesian models, which is to say the probability of an occurrence before the collection of new data. What this means for education is that we need to address not only the effects of automation, but also the very conditions that give rise to these.

Keywords:

• Bayesian statistics
• care
• instauration
• Stiegler
• automated knowledge production
• recursive optimization

After Yang: A meditation on entropogenic exosomatization

The 2021 near-future film, After Yang, written and directed by Kogonada, follows the lives of a family dealing with the aftermath of the death of their adopted android child. In the opening scene, three of the family members steadily come into focus: Jake (Colin Farrell), the hipster-father obsessed with tea; business-mom Kyra (Jodie Turner-Smith); and their adopted Asian child, Mika (Malea Emma Tjandrawidjaja), who is calling out to the person behind the camera: ‘Don’t forget to join us’, she says, smiling.

As the opening credits start rolling, the camera jumps between sets of uniformly dressed families taking part in a virtual dance-off—a mimicking of the choreographed and hyper-synchronized nature of digital lifestyles, and a bleak glimpse at gamification taken to its LARPing logical conclusion. Level 1 passed. Level 2 complete. Nine thousand families eliminated. The intensity is palpable until Jake, unable to fully commit to anything but tea, loses momentum and they too are disqualified without making it to Level 3.

The screen blacks out and only the title comes into focus, auguring the sudden malfunctioning of android sibling Yang (Justin H. Min) in the next scene. Kyra, who is off to work, leaves Jake to deal with the ‘Yang-situation’, as well as with Mika who is clearly distressed. Because the refurbished android store that Yang was bought from no longer exists, and Yang has already started decomposing, Jake is forced to take him to a back-alley dealer who offers to break open his interior core. At first reluctant, Jake agrees to have Yang’s core opened and an electronic chip—seemingly spyware—is found. The dealer, unable to help Jake decipher it, sends him to the technosapien museum curator, Cleo (Sarita Choudhury), who informs him that what he has in his possession is something far more valuable than spyware: a memory bank.

As Jake begins to watch the footage, it becomes clear that Yang did not only have a memory bank, but that he also functioned as the memory bank and affective ‘decoder’ for the other family members. Viewers see, for example, how Yang, now a veritable cultural memory reserve, shares ‘Chinese fun facts’ with Mika. He is also seen sharing moments of intimacy with Jake and Kyra; Jake intimating to him a fascination with the pursuit of the elusive ‘perfect’ cup of tea while Kyra, who hardly displays any emotion throughout the film, unexpectedly laughs out loud at something Yang says.

For me, the film captures something that is far less futuristic than it may at first seem, and I like to think that the philosopher, Bernard Stiegler, would agree. His philosophy, after all, has many of the same leitmotifs as the film. The externalization of memory and knowledge into ‘memory banks’. The forgotten savoirs, or knowledges of how to live, which used to be passed on from generation to generation, but which now is eclipsed by something else, necessitating the need for someone or something to pass on these vanishing intercultural memories as ‘cultural fun facts’. The general malaise—or disaffection, as Stiegler calls it—so prevalent in our societies, the driving force behind the pursuit of elusive affective intensities, now displaced into ‘like’ clicks and infinite scrolling, the magic of life collapsed into the algorithmically choreographed scenes of the digital. The hyper-synchronization of thought packaged as ‘fun’, all the while reducing life to little more than live action role-playing, a virtual dance-off in which the processes of transindividuation are steadily ‘replaced by processes of transdividuation that are’ under the control of corporations who, in turn, are presided over ‘by shareholders who ‘manage’ them according to a single criterion: the increase of dividends—at the cost of psychic and collective disindividuation, at the cost of madness’ (Stiegler, 2019, pp. 46–47; emphasis added). These same themes reverberate in our education systems, the neoliberalization and algorithmic recuperation of knowledge accelerating every year. Like many of the trends that went before it, the latest—the uncritical incorporation of ChatGPT into our teaching and learning programmes—drives not only an unprecedented externalization of our knowledge and memory into these hypomnesic objects, but radically transforms what we understand as knowledge production, now little more than the discovery of patterns and clusters of patterns in data. We have, to be somewhat polemical, been rendered stochastic parrots, Emily Bender’s coinage for large language models—like ChatGPT—that probabilistically stitch ‘together sequences of linguistic forms’ without any real ‘reference to meaning’ (Bender et al., 2021, p. 617) but are believed by many to be displaying some level of theory of mind or consciousness, often helped along by scandalous news reports.

What, then, are we to do about this situation? Is there any room for resistance? After all, as Stiegler reminds us, there is no longer an ‘exit from this situation’—‘the age of technics will never be overcome, contrary to delusions which may be spread far and wide’ (Stiegler, 2013a, p. 34). This does not mean that there is no cause for hope. The point, for Stiegler, is rather that we need to understand not only the new technical milieus we find ourselves embedded in, but also how they are disrupted by novel infidelities which, in turn, disturb our knowledges of life—‘structurally and ever more rapidly’ (Stiegler, 2015). This, then, is the aim of this paper: to think about the organological metamorphosis of knowledge and what is required for us to radically reinvent our politics, such that it goes beyond an elusive search for the perfect cup of tea.

On disadjustment and the transitional object or event

In After Yang, it soon becomes evident that it is the android member of the family who functions as the keeper and transmitter of both knowledge and affect for the unit. For Stiegler, there is a similar displacement of knowledge into digital objects in contemporary societies. He argues, moreover, that the knowledges of how to live—including work-knowledge or skills-knowledge (savoir-faire), conceptual knowledge (savoir concevoir) and life-knowledge (savoir-vivre)—are what ‘constitute spirit’ (Stiegler, 2013b, p. 3) which, here, does not signal some mindless return to religion, but refers rather to ‘the care taken of the objects and subjects of individual and collective desire’ (2013b, p. 6). Stiegler’s most sustained meditation on care is, I would venture to say, in What Makes Life Worth Living? where he ruminates on the transitional object with reference to the work of paediatrician and psychoanalyst Donald Winnicott. Winnicott held that ‘by taking care of her infant, even before the child is old enough to speak’, a mother ‘instils in the child the feeling that life is worth living’ (Stiegler, 2013c, p. 1). Maternal care is thus the ‘transitional object’ that constitutes and conditions the connection between mother and child. The thing about the transitional object, however, is that it does not really exist, even though it takes certain forms: a blanket, a toy, a book. The point is that the form taken by the transitional object opens a ‘transitional’ or ‘potential space’ in which mother and child, in this example, can encounter one another.

In After Yang, tea is the transitional object between Jake and Yang—that which opens an affective space between them. According to Stiegler, the transitional object ‘is the first pharmakon’—that which is both a poison and a remedy at once (p. 2). In Winnicott’s example, the transitional object thus simultaneously provides for mother and child a feeling that life is worth living, making it the ‘point of departure for the formation of a healthy psychic apparatus’ (p. 3), and that which can engender dysfunctional co-dependency between mother and child. The pharmakon is thus also the pharmacological situation: that which at the same time ‘enables care to be taken and that of which care must be taken—in the sense that it is necessary to pay attention’ (p. 4). This is close to George Canguilhem’s understanding of normativity and pathology as these relate to knowledge of life. In the same way Stiegler understands the pharmacological situation as containing both the seeds of poison and cure, so Canguilhem conceives of the organological situation as conditioned by both normativity and the pathogenetic (1978). Stiegler in fact adopts the concept of organology from Canguilhem’s work to indicate the structural shift that takes place when the organic is reorganized by the technical to such an extent that that the external technical milieu can no longer be distinguished from the interior psychic and social milieus. That is to say, the technical, or exterior, now both constitutes and conditions the psychic and social, or interior (Stiegler, 2020, p. 46).

This kind of structural reorganization produces an initial stage of disadjustment which requires processes of transindividuation to be transformed into a kind of temporal ‘rule’ for the metastabilization of the psychic individual as well as the social-technical milieu. As with Winnicott’s transitional object, organological reorganization can either become pathological or produce a new normativity, where the latter would mean that the pathogenic ‘infidelities’ in the milieu have been transformed into ‘elements of meaning, into pathos’ (Stiegler, 2020, p. 93), thereby generating and circulating new affects that can transform the infidelities into a metastable or normative state. Such an initial state of disadjustment is provoked, in the film, by Yang’s malfunctioning which triggers a shift in the structural organization of the family. Similarly, pedagogical milieus are now in stages of disadjustment which require that the uncomfortable—even painful—intensities of the world be embraced, and that pedagogues simultaneously recalibrate classroom activities and academic narratives away from the mindless externalization of knowledge and knowledge production. For while it may seem to some that ChatGPT and its image counterparts—like DALL-E, Craiyon, Midjourney and Stable Diffusion—are little more than distractions or entertainment, perhaps even useful educational tools, it seems to me that their value lies really in the fact that they are reflecting our collective consciousness back to us. Our externalized memory and knowledge are, after all, precisely what they are trained on. Yet, it is not quite our collective consciousness—not in the ways it was initially produced. Instead, it is an algorithmically modulated version thereof—and this version has no time for the process of making tea, for watching the tea leaves dance in the water as their flavour seeps in. No, this version is the trickster, the hustle, the short-cut; in a word, the end product. It’s the version where the actual processes of production are eclipsed—and this requires far more than AI ethics because, as Benjamin Bratton and Blaise Agüera Y Arcas argue, ‘the “ethics” discourse doesn’t go nearly far enough to identify, let alone address, the most fundamental short-term and long-term implications’ of the kinds of ‘cognitive infrastructures’ we are embedded in and confronted by (2022). So before we can truly grapple with the problem of ethics, what is required is a demystification of the processes by which knowledge and knowledge production become subsumed under the rubric of recursive optimization for predictive performance and, concomitantly, what it means to be a ‘hyperindividuated yet networked’ self (Halpern, 2014, p. 138). To this end, I provide a brief overview of what I call the ‘Bayesian model’ of being, knowing and learning.

The Bayesian model of being and knowing: A generalized denoetization

If a Bayesian calculus could be said to undergird at least part of the problem of the hyperindividuated but networked self, the conditions that make this calculus possible are no doubt numerous though four, in particular, stand out in my mind: extraction, surveillance, addiction and the use of Bayesian statistics. The extractive logic of our algorithmic ecologies has been well documented, though the methods for gathering our data are continually aimed at gaining new ground. To this end, our biometric information—including facial expressions, voice data and gestural samples—are acquired from our technological devices, often without our consent, just like data of Yang’s family were acquired by the company that manufactured him via his memory bank. The education sector, too, is replete with this kind of data-centrism, driven by extraction logics that propagate reductive ideas about learning and teaching as data problems to be solved algorithmically. We need think only of year-end feedback reports on classroom experiences or performance appraisals that treat relationships as numbers and other quantifiable variables. This is a logic, as Kate Crawford argues, that can be summed up in the following byline: ‘you shall know them by their metadata’ (2021, p. 185). This is not to say that these practices are not contested—they are—but effective resistance becomes harder each year as surveillance technologies develop and spread. Besides, these technologies are deeply connected to processes that encourage addictive and obsessive behaviours because, far from simply being aggregated data-points, personalized advertising and social media are designed to induce technological overdependence by creating dopamine loops to keep us clicking and scrolling and, in so doing, increase our online participation data footprint. The problem, it would seem, is no longer just about the domestication of desire via the family, State, education or capitalism; it is now also about the compulsive and habit-forming dynamics induced by the algorithmic architectures we find ourselves embedded in (for a longer discussion, see Gray 2024).

This kind of overdependence on hypomnesic objects is so well portrayed in After Yang. There are numerous examples to choose from, but what stands out for me is the exosomatization of parenting skills (savoirs) to the android sibling, Yang, and the concurrent impulses fed by the continual externalization of the self to compensate for ‘the retentional finitude of the organism’ (Hui, 2019, p. 202). For Kyra and Jake these pathologies manifest as work addictions and tea-related obsessions. Interestingly, the term Internet addiction first began to circulate in 1996, though the indicators used for the diagnosis thereof were developed with reference to alcohol addiction as it was then captured in the Diagnostic and Statistical Manual of Mental Disorders (DSM). As with any addiction, these new computational addictions were allied with dopaminergic ‘highs’—not surprising as dopamine is correlated with feelings of well-being (Carstens & Gray, 2023, p. 12). But dopamine is itself a pharmakon: when enough is released into the body, positive psychological and physical effects are felt; too little and depression seeps in; too much and it leads to agitated or even aggressive behaviour, poor impulse control and interrupted sleeping patterns. The introduction of ChatGPT and its counterparts are, similarly, prompting dependency behaviours in learners who, rather than enjoying the process of learning, are tempted by the easy-fix knowledge production methods these kinds of models offer. Needless to say, there are many drivers behind these kinds of dopamine and adrenaline seeking behaviours, but as many algorithmic processes remain opaque it is worth taking some time here to unpack at least one way in which these models are used to create compulsion loops for internet users.

The history, it should be remembered, is neither linear nor one-dimensional, some of the lines leading back to Charles Babbage and Ada Lovelace, others to George Boole, George Cantor and Richard Dedekind, others yet to Alan Turning, and many crisscrossing between. Of late, many scholars have also genealogically traced our current algorithmic societies to the inception of cybernetics (see, for example, Hayles, 1999). One of the early adopters, Frank Rosenblatt, argued—in what can only be called visionary clarity—against a focus on the symbolic and linguistic manipulation of symbols so in vogue at the time. The problem with this line of thinking, he held, is that it fundamentally disregards the notions of scale and emergence. Instead, he ‘proposed that learning, whether in non-human animals, humans, or computers’ be ‘modeled on artificial, cognitive devices that implement the basic architecture of the human brain’ (Halpern, 2023). To put it differently, Rosenblatt argued that instead of focusing on how humans use and interpret abstract symbols, the focus should shift to a networked approach that considers learning in terms of probability. But before this idea could take flight, in part through Bayesian statistics, another element would need to be added: belief in the computer, despite the opacity of its operations. This belief would, moreover, would have to exceed belief in ‘human knowledge’ for a ‘proof’s specifics’ (Joque, 2022, p. 64). One event that contributed to this transformation took place in 1976, when Kenneth Appel and Wolfgang Haken announced that they had solved the ‘four-colour theorem’ using a computer. The theorem, in short, asks whether it is possible to mathematically prove that the regions of any given map can be coloured in with no more than four colours, the rule being that no two adjacent regions may match in colour. With the help of a computer, Appel and Haken were able to reduce all possible maps to 1,936 configurations. ‘If those maps could be colored with four colors, they argued, then so could any map’ (p. 62). The importance of this feat was not only that it facilitated a new kind of belief, in the computer itself, but also that it augured a new form of hypomnesic memory—what would, roughly forty years later, become what Stiegler calls digitized hypomnesic memory, or what might even be called probabilistic hypomnesic memory.

Although Bayesian statistics has not fully replaced frequentist statistics, the dominant statistical paradigm of the twentieth century, it does mark an important shift. Frequentist statistics is well known. Simply, it is a method that treats ‘probability’ as ‘frequency’, meaning it draws conclusions from sample-data about the frequency of a statistic in the long-run. For example, if after executing a hundred coinflips, a coin lands on tails 40% of the time, we would say that the probability of the coin landing on tails is 40%. What is being measured here, in other words, is the probability distribution or frequency of a statistic (Joque, 2022, p. 147). If, however, it is already assumed that the coinflip is fair, the model could be adapted to also measure the probability distribution of the statistical model itself—and this, precisely, is what Bayesian statistics do. So, unlike frequentist statistics, Bayesian probability does not start by assuming the truth of a null hypothesis, such as that a coinflip is fair, but instead understands probability ‘as a measure of subjective belief’ (p. 148). Bayesian statistics thus lends itself to updating beliefs about the world as more data—which amounts to ‘evidence’ here—is gathered in real time. But here we find the caveat: Bayesian approaches use prior distributions—or knowledge, or conjectures, or beliefs—to quantify parameters and weighting distributions. What this means in practice is that any search undertaken on Google, for example, aggregates information for a user to see based on their own priors, or online behaviours. To give an example, if a user searches for something like ‘flat earth’ on Google, and most of their searches show that what they have clicked on confirms the view that the earth is indeed flat, their search results will be algorithmically slanted towards more results that confirm this belief. In a sense, then, Bayesian statistics provides the ‘mathematical framework in which evidence can be transformed into belief’ (Joque, 2022, p. 161). More pernicious, as we see every day with fake news, disinformation, misinformation and conspiracy theories, is that this transformation of data-evidence can bring about real changes in the beliefs of an individual or group despite, or perhaps because of, weak priors—and ignorance of weak priors. In a sense, we are all functioning like Bayesian models, which basically translates to: If your prior is Alex Jones, your posterior probability is sure to be a wholesale denial of the Sandy Hook Elementary School shooting. The take-home message, then, is: Update your priors! Jokes aside, this poses real problems for both speculative philosophy and education. So how might we develop tools aimed at priors rather than outcomes because this, I would argue, is where the short-circuiting that Stiegler talks about occurs.

The instaurative gesture: Between organological normativity and pathology

Contemporary life has, to a large extent, become characterized by generalists who have to continuously adapt to changing conditions so that uncertainty ‘becomes a primary affective orientation’ (Dixon-Román & Puar, 2021, p. 7). Although uncertainty is ‘embedded in the calculus of statistical probability’, it is not so much about the ‘control of the emergence of the uncertain’ as it is about creating and directing that uncertainty (p. 7). It should be said though that uncertainty per se is not the problem—any milieu is marked by infidelities and phase shifts, neither of which are absolute markers of normativity or pathology. As Canguilhem points out in On the Normal and the Pathological, because we use these concepts too narrowly, we misunderstand—and mis—the pharmacological dynamics between the two terms. As he says: ‘the pathological state can be called normal to the extent that it expresses a relationship with life’s normativity’, where normativity simply expresses a certain relationship between a being and the ‘fluctuations’ of that being’s environment (Canguilhem, 1978, p. 137). Cure, or normative health, from this point of view, is simply ‘the reconquest of a state of stability of physiological norms’ (p. 137), meaning the more a system is able to deal with change, and thereby return to what Canguilhem’s student, Gilbert Simondon, would call a ‘metastable equilibrium’ (2020: 235, emphasis added), the healthier it is, whereas an inability to deal with disadjustments and infidelities in a milieu will signal pathogenic adaptations, which Stiegler distinguishes from normative adoptions.

Adaptation refers here to a kind of ‘hyper-acculturation’ to a transitional event which, in its turn, triggers processes of disindividuation: a premature interruption in processes of individuation and transindividuation. This causes the metastable state to attune to certain basins of attraction in the region of a phase space, where a phase space constitutes all the possible states of a dynamic system. Said otherwise, singularities are distributed as intensive attractors or tendencies in the pre-individual ‘fund’ from which the world of fully formed subjects and objects unfold in extensity. This takes place through ‘progressive differentiation’ prompted by a ‘cascade of symmetry-breaking events’ (DeLanda, 2002, p. 72). For example, H₂O can be either liquid (water), gas (vapour) or solid (ice). In this example, the attractors are the regions that are defined according to the conditions of temperature, such that when a certain threshold in temperature is reached, the trajectory of the system tilts asymptotically to the basin of attraction in that region of the phase space. These tendencies are neither inherently ‘good’ nor ‘bad’—they simply establish a pharmacological situation that can tilt to normativity or pathology depending on the context requirements and which of the immanent conditions are actualized.

As with H₂O, the phase spaces of human psyches are marked by singularities—many more in this case—based on experience, genetic make-up, hormones, current conditions, and so on. When these singularities act as attractors for a person’s trajectories, they may influence the behaviour of that person. In Japan, for example, hikikomori, which refers to people who have withdrawn from society for a lengthy period, was first observed and reviewed by Tamaki Saitō. In his book, Hikikomori: Adolescence Without End, he describes how psychosocial suffering begins to manifest in behavioural patterns like apathy, truancy, anthropophobia, agoraphobia, regression, suicidal ideation and obsessive-compulsive behaviour, though these were always accompanied by a protracted withdrawal from society (Saitō, 2013, p. 11; pp. 30–53). What Saitō also noticed was that most hikikomori tended to be male, first-born and from a mid to upper class background. As might be expected, they were also generally well-behaved—or had been until the withdrawal—though they were often maladjusted or alienated members of society (for a longer discussion, see Gray & Eloff, 2023). In Stiegler’s parlance, pathological adaptation has taken place, which he distinguishes from adoption, where the former generally refers to a botched individuation process whereas the latter signals the production of new singularities in the phase space. To go back to Winnicott’s example of the transitional object, it helps to draw here on a related argument made by Alexander Krieg and Jane Dickie that ‘securely attached children function well because they balance proximity seeking and exploration’ (2011, p. 62). In other words, securely attached children have developed the ability to feel safe during short periods of detachment or separation—for example when going to a playmate’s house or attending classes at school—because they have learned how to transfer that feeling from primary to secondary caregivers. That is, the transitional situation has been normatively adopted, which is to say an ‘enrichment’ rather than ‘a restriction of the possibilities of the individual’ (Stiegler n.d.) has been produced. When children do not develop this ability normatively, perhaps because of an absent and detached father like Jake in After Yang, the negative attachment and detachment methods learned begin to form basins of attraction that condition the later life trajectories of the person. A different way of saying this is that these basins of attraction form the priors on which algorithmic models thrive. Transposed to education settings, it becomes clear that this applies also to knowledge and knowledge production. For instance, when a student uses ChatGPT to produce their class assignments or research projects and they get away with it, the algorithmically modulated version of what it feels like to produce knowledge begins to function as an attractor. The student has, however, learned nothing about the actual process of knowledge production and the ‘high’ of the ‘a-ha’ moment researchers experience. All that has been learned is what recursive optimization for predictive performance feels like—what it means, in fact, to have become a Bayesian model of being, knowing and learning.

I want to suggest here that in order to develop speculative tools aimed at priors rather than outcomes, we cannot think of digitality solely in terms of alienation, exosomatization and other poisonous pharmaka. Moreover, if normative health, as Canguilhem and Stiegler argue, is about the ability to deal with change in such a way that adoption rather than adaptation takes place, then it seems to me that Étienne Souriau’s concept of instauration might be of use. Originally from his 1939 work, L’Instauration philosophique (1939), translated as The Different Modes of Existence (2015), Souriau creates the concept of instauration because ‘terms like production or creation’ remain for him, in the end, ‘too ambiguous’ (Lapoujade, 2021, p. 51). Instauring is in some ways quite close to the idea of individuation as it is an insistence that each existence proceeds not from a creator but from a gesture or ‘arabesque’ that is immanent to life (p. 5). The instaurative gesture, like the individuation process, is a response to a problematic and thus unfolds with it a ‘new organization of existences’, each of which has its own ‘network of relations’ that is gradually established, such that they ‘mutually limit and consolidate one another in a cosmos’ (Lapoujade, 2021, p. 53). But although there are overlaps between individuation and instauration—and it could perhaps even be argued that what Souriau was trying to create was something like a theory of individuation—there is something unique about Souriau’s concept that makes it other than individuation, and that is his argument for the richness that emerges when ‘several kinds of existence’ are accepted as fully existing (Souriau, 2015, p. 101). That is, he argues for an acceptance of what his interlocutor, David Lapoujade, calls the ‘lesser existences’, such as ‘reflective existence’ or ‘spiritual existence’ alongside biological existence (p. 101). For Souriau, an acceptance of these lesser existences as fully real depends on whether we think the verb ‘to exist’ has exactly the ‘same sense in all of its uses; whether the different modes of existence … all deserve the title of ‘existence’ in full and equal measure’, be that Leibniz’s monads, Heidegger’s Dasein, or Deleuze’s difference (p. 103). The problem of existence thus requires, according to him, an understanding of three aspects, the first two of which have actually been discussed in this paper all along, namely: 1) the intensive modes of existence; and 2) the connections and shifts between different modes of existence, and the conditions by which we might pass from one to another, which is to say the conditions that give rise to phase shifts (p. 106).

In the context of digitality, and read in the context of normativity and pathology, instauration, I would argue, signals the implex state between adoption and adaptation, a gesture that refuses a paralysis of the spirit. In this lies the power to unfold new organizations of relations that actively create and propagate novel methods of care so that ‘every being’ is ‘instaured, the soul as well as the body, the work of art as well as the scientific existent, an electron or a virus’ (Stengers & Latour in Souriau, 2015, p. 21). The point is that instauration, as a progressive unfolding, does not aim to establish in advance where to arrive and ‘and then mobilize the means by which this end might be realized’ (p. 80)—as would be the case with knowledge production via ChatGPT. Instauration, instead, allows for the process of gradual unfolding—for care to be taken as new relations are built and adopted.

In conclusion

The question of how to do academia differently is surely on the minds of many pedagogues, caught as we are between cycles of competitiveness, neoliberalist austerity, inconsequential but growing numbers of administrative tasks, learner (and sometimes also teacher) apathy, and burnout and over-medication, to name just a few of the symptoms. ‘All this belongs to the process of grammatization as the exosomatization of the life of the mind and spirit’, as Stiegler says, ‘which, set outside itself, interiorizes itself by already dreaming of the next stage of its self-exosomatization’ (2019, pp. 109–110). Perhaps this can be summed up as carelessness. If we are to learn how to dream again—to fabulate futures hitherto unthought—we must begin to collectively creative new methods of care: speculative tools aimed at priors rather than outcomes, so that we may readjust to the ‘real trauma that afflicts our epoch, and that provokes all the others’ (Stiegler, 2020, p. 9). These traumas, as I see it, cannot be addressed solely at the level of actualization—there must be an attempt to counter-disrupt the disruptive logics at the heart of digitality, to change the basins of attractors that propagate carelessness and disaffection, and lead to pathology rather than normative health. Interestingly, Canguilhem argues that normative health is not so much about not falling sick at all as it is about the idea that normative health must contain ‘the power and temptation to fall sick’ (1978, p. 117). This is because without this power there is no possibility for the posing of a problem.

In his work, Simondon argued for the existence of a pre-individual field or fund from which life unfolds into extensity, as well as for the replacement of ‘the notion of stable equilibrium with that of metastable equilibrium’ (Simondon, 2020: 235, emphasis added). Instead of being viewed in terms of the maintenance of homeostasis or adaptation to environmental changes, a dynamic system or living being is therefore understood as being in a perpetual motion of internal tension that creates ‘problems’ to be resolved at local levels, for example regulating heat or energy expenditure. This tension is important because it produces a problem of disparation to which a system or organism can respond through normative adoption or pathological adaptation. The relative ‘solutions’ to a problematic—whether organic or organological—forms part of a schema of ­differentiation and integration, or ongoing, partial and relative processes of individuation and transindividuation. On this account, a living being is ‘both more and less than unity, conveys an interior problematic and can enter as an element into a problematic that is vaster than its own being’ (p. 8). So, the individual is a ‘being that translates potentials which are incompatible with one another into metastable equilibria that can be maintained by means of successive inventions’ (p. 239, emphasis added). Basins of attraction—or priors—are generally resistant to change because they have been iterated many times over. They are thus long-term trajectories. When a problem presents itself, a person’s psyche naturally tends to want to solve the problem with recourse to existing attractors, not only because the trajectory of any given system automatically tilts asymptotically to the basin of attraction in specific conditions, but also because this seems less entropic to the psyche. This is because less ‘work’ or energy is needed to solve a problem if it can slot into an existing pattern. For Stiegler, digital automation has become such a large-scale basin of attraction in contemporary societies. To be clear, automation is not a problem per se. ‘A biological cell, for example, is a sequence of instructions and this sequence of instructions is automatic. The reproduction of life is automatic. When you have something that is not automatic, it is a mutation, which produces a monster’ (Stiegler, 2015, p. 16). The issue, then, lies rather with the way in which digital automations ‘short-circuit the deliberative functions of the mind’ (Stiegler, 2016, p. 25), thereby instantiating dividualizing and disindividuating processes. For Stiegler, unlike Simondon, this is not a neutral operation that forms part of all individuation processes but implies, on one level, the short-circuiting of ontogenetic and sociogenetic processes and, on another level, the proletarianization of knowledge. This proletarianized form of knowledge—and now also of knowledge production, as I have argued—becomes the attractors or priors of our collective trajectories.

In closing, then, it seems to me that the political project now is about how to construct and prefigure—instaure—alternative modes of existence, including modes of knowing, that have enough consistency to replace the viscosity of Bayesian logics—as well as every other intersecting line of domination. This is no small task given the complexity each person is tasked with every day, combined with the lack of shared practices of meaning and care—pathos, as Stiegler understands it—to ameliorate the sense of an ending. ‘From incidents to accidents, from catastrophes to cataclysms, everyday life has become a kaleidoscope where we endlessly bang into or run up against what crops up’ (Virilio, 2005, 3). And so it is that ‘in this broken mirror, we need to learn’ clearly how to ‘make out what crops up more and more frequently and, more to the point, more and more rapidly’ (p. 3). The issue at hand, in other words, is one of care: how to take care of the pharmacological situation, how to take care of each other, how to take care of how we learn and produce knowledge and, in this final hour, how to take care of the world.

Competing interests to declare

The author reports there are no competing interests to declare.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Chantelle Gray

Chantelle Gray is a Professor in the School of Philosophy at the North-West University whose interests span critical algorithm studies, queer and critical feminist theories, experimental music studies, pedagogy, anarchism, and Continental philosophy. The interdisciplinary nature of her research allows her to ask critical questions about how to take care of humans, technologies and ecologies in the digital age. She is the Chair of the Institute for Contemporary Ethics (http://contemporaryethics.org/), the co-convener of the South African Deleuze & Guattari Studies Conference, and an editorial board member of Somatechnics. Her books include Deleuze and Anarchism, co-edited with Aragorn Eloff (2019, Edinburgh University Press), and Anarchism after Deleuze and Guattari: Fabulating Futures (2022, Bloomsbury).