Challenging Logistical AI

Politicizing the rise of Artificial Intelligence while autocrats are gaining momentum, the Ambient Revolts conference moved onto new ground: ‘Logistical AI’. It is new ground insofar as it has hardly been covered by academic or journalistic knowledge production. Of course, there is a lot about AI in general and there is also a lot about logistics in general, but there is hardly any literature about the intersection of these terms that I propose calling Logistical AI. So, out of urgent necessity, in my view there is a need to invent a critical discourse on Logistical AI. In order to do this I first sketch the emergence of Logistical AI as a field of politics, then introduce the seemingly unrelated work of Sandi Hilal and Evelina Gambino within this emerging field. Finally, I reflect on the struggle within and against Logistical AI as a politics of Artificial Artificial Intelligence (AAI), raising critical issues of agency and labour.   

First, the current stage of capitalism. As many have convincingly argued, production is ceasing to be of importance and circulation is becoming the only game in town. Logistics is the hidden force behind this tendency. AI in turn promises to be the perfect technology for this stage of ‘frictionless’ and ‘seamless’ capitalism focused on circulation. Computer programmes are supposed to generate their own rules and engender their own programmes and outcomes, seemingly keeping things in ‘continuous motion’ and ostensibly ‘creating something out of nothing’.

Second, consider what the big players of the ‘Partnership on AI’ are up to – IT giants like Amazon, Google, Facebook, etc. They are investing tremendous amounts of resources in AI to manage transnational corporate empires more efficiently, with new user applications by-products of this primary goal. This, of course, is a matter of logistics. Why? Look, for instance, at the publicly staged step the IT giants are taking to become the architects of the so-called Smart City. This step entails the challenge of managing the circulation of goods, people, data and capital in logistically intelligent ways.

These are just two examples that show that Logistical AI is happening. What is important for this discussion is that, just like the fantasy of frictionless circulation, the increasingly privatized form of AI-driven city governmentality is meant to magically neutralize pressing social and economic frictions along the way. For instance, as Francesca Bria’s and Evgeny Morozov’s work on the politics of the Smart City suggests, smartness – as an indicator of frictionless circulation – is gradually overwriting politics. To challenge Logistical AI the question of politics must be reintroduced, less through a totalizing approach but rather from specific perspectives – sideways.

Some important impulses to politicizing Logistical AI come from political geography. For instance, Louise Amoore’s work on (‘self-learning’) algorithms in the context of mobility and capital circulation, and Deborah Cowen’s work on the securitization of transnational supply chains. Both point to a major but often neglected paradox in the governmental ambition to optimize the circulation of capital. As this ambition entails a securitization of supply chains, e.g., the transformation of ports into security zones, the governmental claim is that all of this is being done in the name of ‘national security’. But doesn’t the securitization of circulation threaten national security and vice versa? In other words, if you render incessant circulation securable – enabling a constant and frictionless passing of nation-state borders – how can you at the same time foster national security that traditionally hinges upon containment? Isn’t this an insurmountable paradox? Cowen shows in her book The Deadly Lives of Logistics that both circulation and security are actually possible at once, because within logistical landscapes the logics of circulation and security are recalibrated to that end. Cowen writes: ‘The stretching of logistics systems across [national] borders into “pipelines of trade” means that supply chain security recasts not only the object of [national] security but its logics and spatial forms as well.’

That AI-driven governmentality reconciles circulation and security is also apparent in the deployment of AI at state borders. Amoore notes in her book The Politics of Possibility: ‘in order to learn, to change daily and evolve, [algorithms] require precisely the circulations and mobilities that pass through.’ This observation, as she also makes clear in an interview I have conducted with her, is part of her larger Foucault-indebted argument about how governmentality is less concerned with prohibiting movement than with facilitating movement in profitable ways. The role of ‘self-learning’ algorithms seems to be significant in this context, since – like capitalism – they are produced by and productive of movement. Thus the very securability of movement and circulation in general becomes the condition for profitability within an emerging logistical infrastructure space that architect and urbanist Keller Easterling calls ‘Extrastatecraft’. The thirst for traffic is the common denominator of self-learning algorithms and circulation-based capitalism, as well as of logistics and security.

In view of this, it is becoming crucial to investigate the connection between seemingly unrelated things such as drone wars – an extreme example of AI-driven security – and refugee escape routes. After all, as a workshop at the Tacit Futures conference explored, they both rely on the (public-private) infrastructure that is increasingly becoming a logistical matter of data modelling and computer programming. That said, it is becoming ever more important to expand the emerging field of Logistical AI in migration research. There are traces of this in the work of the scholars who jointly authored the study ‘Logistische Grenzlandschaften’ (published in English in a special issue of The South Atlantic Quarterly). Manuela Bojadzijev, Sandro Mezzadra and their colleagues analyse the ongoing logistification of migration management. An emergent governmental strategy – or should we instead call it a fantasy? – has arisen. Whatever it is, the logistification of migration management wants mobility to become ‘a programmer’s game’, as Brett Neilson notes. As a result, it wants migrant labour to arrive at its desired destination as efficiently as Amazon items. The logistification of migration management is a dangerous tendency that should not go unnoticed or unprotested.

The refugee camps in the occupied territories formerly known as Palestine are circumscribed by what Israel cultivated in the last two decades as the ‘most extreme examples of privatized security in the world’, as economic theorist Shir Hever puts it. Moreover, Israel, as a neo-colonial state, has nurtured an IT sector that has developed many of the post-9 / 11 security technologies. While Israel’s economy is the ‘most tech-dependent in the world’ (Business Week), supplying the international market with security technology, is IT sector catalyses and accelerates the transformation of the region into an experimental laboratory for Logistical AI. This unsettling marriage between circulation and security ‘inspires’ political and economic actors around the world. What Israel pioneered in the 1990s as a high-tech marriage between security and circulation-based capitalism, many embraced in the post-9 / 11 era.

The post-9 / 11 new normal likely has distracted many observers from the fact that Israel’s economic model undermines the peace process in the region, as Naomi Klein convincingly argues. Instead of committing to the peace process, Israel manages the region as a high-tech security landscape where all natural and built features serve the ends of militarized movement control, as architect Eyal Weizman suggests in his book Hollow Land (2017). We are talking about a neo-colonial battlefield that is not temporary but permanent, engendering spaces of inhabitation over time. This permanent battlefield and the spaces of inhabitation that are superimposed on top of it, eventually both need to establish interrelated supply chains in order to sustain themselves as such, so that the logistics of war and the logistics of everyday life overlap each other, becoming interdependent and indistinguishable. The logistical nightmare echoed in others on this planet is cause for further study.

The features of mobility security are an assemblage of structures that organise the space and time of the region. This is done in a manner so violent and so alien that it is hard to imagine that human designers have been responsible for it. It is easier to imagine AI at work: cold, detached, far from human suffering and politics. Just like black-boxed self-learning and self-reproducing AI technologies seem to be. Yet, even if this image of a non-human actor may help to picture the horror, it is important to remember that all AI is largely human and political by design. If the regime of movement control is giving itself a distinctly non-human face – then this suggests that an unaccountable power is responsible for the violent design. This clearly depoliticizes the situation. Just as refugees, whose movement is meant to be as constricted as possible, are denied their status and potential as political subjects. Yet, Hilal's work as an architect, researcher and initiator of many seminal projects on decolonization shows the contrary: refugees are the most important political actor in the region. After all, it is their existence, their doings and their organising that destabilize the neo-colonial security matrix by consistently calling the occupation and its regime of movement control into question.

As one of the refugees from the camps in which Hilal has been active, says, and I am paraphrasing from her book Architecture After Revolution: the better organized our camp becomes, the more effective our struggle for decolonization. Better organized camps as a product of alternative logistics of refugees suggest a grassroots counter-politics capable of challenging Israel’s highly securitized logistical landscape. This said, the political potential of refugees comes into view anew in the face of other logistical landscapes inspired by Israel. For instance, the designated ‘reserve army’ of ‘flexible labour’ has become the ‘raw material’ of the logistification of migration management introduced in Europe after the Summer of Migration in 2015, when millions of asylum seekers entered EU territory. If this entails a renewed recalibration of the relationship between (nation-state) security and circulation-based capitalism, then this process – overshadowed by rising right-wing populism – last but least puts the political potential of refugees up for a seminal test.

Let us turn now to the other site of reflection: Georgia. This country is at the centre of ambitious infrastructural investments aimed at transforming it into a logistics hub for the Chinese-led New Silk Road project. These developments are reshaping Georgian territory and its economy. There are reasons to consider this process a logistics revolution. Some aspects are of great relevance to the discussion of Logistical AI. To begin with, geopolitics has defined the territorial competition as a ‘struggle for the heartland’; as Gambino’s work suggests, the logistical power of the New Silk Road introduces a new heartland. This new heartland is, quoting from Gambino’s work, ‘a deterritorialized cyborg, resulting from the agglomeration of infrastructure, territory, manpower, and resources’. And it is this ‘deterritorialized cyborg’ that should be explored: how does its source code look? How does its technological design work? What is its promise of seamless circulation all about? How does it integrate security designs? Moreover, despite the advancement of increasingly AI-driven automation, the logistics sector still relies on great amounts of labour power. This in fact also holds true for AI in general, which is why the term Artificial Artificial Intelligence has been coined. It suggests that AI has to hide its dependence on human labour in order to appear magically autonomous. Now if the cost of this labour and its bargaining capacity are the chief obstacles to seamless circulation – then what possibilities do we face today to disrupt seemingly fully-automated supply chains and logistics networks?

Prior to her work on logistics in Georgia, Gambino took part in the creation of a network of precarious workers, migrants, researchers, grassroots unions and activists. One of the imperatives of this unique network is that one’s subjectivity does not exclude the other. She has been realizing bottom-up initiatives partly in highly secluded migrant labourer settlements in Italy, including co-research and collaborative educational projects documented in her article ‘The Gran Ghettò: Migrant Labor and Militant Research in Southern Italy’ in The Borders of ‘Europe’: Autonomy of Migration, Tactics of Bordering edited by migration researcher Nicholas De Genova. What Gambino calls ‘collective militant research’ is an attempt to develop a research practice from within an emerging struggle, seeking to build an understanding of research as integral to the composition of a class consciousness, rather than outside or above it. The possibilities opened up by this attempt, as well as its limits, inform her current work on logistics.

Taking all of this into account, we can firmly grasp the common denominator between Hilal’s and Gambino’s tremendously inspiring work: in their highly politicized investment to research, both of them show the dispossessed and invisibilized actors of the increasingly AI-driven circulation management regime are political subjects in their own right arising from within the tectonic violence of contemporary logistical landscapes. They are actors who coin new, seminal forms of agency by resisting submission to the power of Logistical AI.

The challenge involves grasping the political consequences of mobility policies driven by the logics of AI-driven goods circulation under the motto: ‘let us apply what, for instance, Amazon has mastered in the field of goods to the domain of human movement.’ Applying the logistical designs of goods circulation to the domain of human movement means applying AI to the managing mobile labour. It is a dangerous fantasy. More than that, this impertinent idea is being practically implemented at various levels of governmentality. In reading Gambino and Hilal, one can see how Logistical AI does not simply replace human labour, it reframes and re-narrates what labour is supposed to be. Logistical AI engenders new labouring bodies. Retrieved from the ‘surplus population’, these bodies are coded as ‘machine spares’ or ‘hardware periphery’. Humans are not replaced by AI but repurposed as replaceable modules of AI and subordinated to its post-human logics.

At this juncture, two politics of labour with respect to AAI come to the fore: 1. that which is supposed to render logistical infrastructure frictionless, enabling seamless circulation of not only goods but also of labouring bodies; and 2. that which is channelled through logistical infrastructure and supposed to arrive just-in-time, and also, profiled via algorithms, to fit in seamlessly in the workplace. Reading Gambino’s research as an intervention into the former and Hilal’s into the latter, both gesture towards a politics of AAI that is about the hidden labour of workers becoming ever more invisibilized by AI-driven governmentality. Their research intervenes in logistical AI because it opens up space to think about how to struggle within and against this form of power over labouring bodies – from the point of view of these very labouring bodies. How to organize and struggle within and against logistical power is a question that implicates the most precarious actors in the AI-driven circulation regime: refugees, migrant workers, day labourers, etc. How are they not just instrumentalized by but actually subverting this rising form of infrastructural power? The politics of AAI is then brought to the fore: human labour that is invisibilized within increasingly AI-driven logistics.

This in turn triggers a number of other questions. If there is something like an autonomy of migration (that counters the fantasy of AI-driven movement control) – what do the invisibilized as political subjects to come allow us to explore and realize about this autonomy in the context of Logistical AI? If logistical power is all about organizing the movement of things as seamless and virtually endless circulation – then what does it mean to challenge that power by making visible the systematically invisibilized labour of the invisibilized? What, if anything at all, becomes visible in acts of refusal and disruption, as suggested by Gambino? Or in acts of pausing, resting and hosting, as suggested by Hilal? And why and how is it actually possible to reclaim political agency along the way?

It will become ever more important to inquire after ‘how logistics routes provide new connections, and somewhat unexpected transnational possibilities for cementing existing alliances’, as Gambino suggests. Thus, we are also able to build on an idea Cowen discussed about at the conference ‘Friendly Fire’ and in an interview before the talk:

The emphasis on circulation in logistics systems gives a special power to the act of disruption [...] [such as] the blockade or the occupation [...] This is not only because of the immediate effect of disruption, but also because of the space of the convergence itself, and how alternative relations of care and provision – alternative logistics – anchored in relations of reciprocity and solidarity can emerge through acts of disruption.

Reflecting on and articulating this vision will hopefully inspire new alliances. Last but not least, this will hopefully also be heard as an urgent call for further inquiries into Logistical AI, especially by foregrounding the often neglected politics of AAI that, for example, reintroduce issues of labour where labour is ostensibly no longer relevant.