ARTIFICIAL INTELLIGENCE AND POLITICAL PARTICIPATION IN THE SMART CITY

eeefff (Nicolay Spesivtsev and Dzina Zhuk)

Text was written for AMBIENT REVOLTS · How Can We Rethink Political Agency in an AI-Driven World? conference in 2018

The urban fabric we inherited from the pre-Internet age, today is reconfigured by computation-based economies. This transformation gives rise to the many effects which require a reinventing of a critical language: citizens are pushed into the position of users of the city, privatization of the infrastructure, a view of the city from the perspective of efficiency – all that are the results of a shift in understanding of the processes of computing systems development that are installed into the urban fabric and are intruded to its inhabitants. This state of affairs dictates a situation in which it is impossible not to be a user. At the same time we can no longer discuss users of what we would like to become , as this situation is imposed on us without our request or approval. The forces hiding behind adding the prefix “smart” to the city minimize the risks of their investment projects, reconfiguring the political field in a convenient way for them.

The experience of physical presence in the city today does not give a complete picture of the processes in which a contemporary citizen is involved. A person immersed into the urban environment becomes not only a user, but also a “sensor” radiating data which participate in feedback loops of intelligent systems installed into urban tissue. These feedback loops are going through our bodies in an adaptive way, become parts of our biography and entangle us with non-human life in a soup of extracted data, including animals, any sorts of machines, city geology and climate. The critical approach to the notion of smartness for us is built on these feedback loops, our relations to them and the possibility to contest rights to change them. This approach also focuses on algorithms that get data from feedback loops on their inputs and adapt reality according to political interests of parties controlling these algorithms.

Intelligent machines

Against this backdrop we propose to look at how an algorithm-driven city landscape is being constructed. For this purpose we need to unbox the term artificial intelligence (AI) and in the course of this shift the focus on intelligent machines because they are concrete implementations of the general idea of AI installed into the techno-socio-political landscape. The precise look at intelligent machines gives us the possibility to approach complex assemblages of digital infrastructure, on the one hand composed of data-sets, sensors installed in the cities and algorithms, on the other hand composed of IT- workers that create, teach and tune these machines.

When we talk about IT-workers that are engaged with the production of specific intelligent machines, we refer to the contextualization of these collective knowledge production methods with all its restrictions and contradictions. We also refer to the specific algorithms that form these intelligent machines and furthermore perform knowledge production. The results of these processes include the transformation of material by other machines (as in the field of generative design or architecture) or the modulation of human relations (as, for instance, inside of the service of ordering a taxi, where intelligent machines coordinate users and drivers’ movements in a profitable way for the company).

Our interest is the gap – respectively the intersection – between the algorithm as a form of mathematical modeling, the social imagination and the very specific implementation of algorithms in social, political and economical structures.

Algorithmically Mobilized Labour Force

Holding these inputs in mind we want to rethink the traditional tools and methods of claiming the right to the city. When confronted with the “smart city”, we need to reinvent political instruments and a language in which to negotiate. A language that would politicize the engineering discourse that dominates now in urban governance.

In order to do this, we we propose to focus on what the so called smart city does to human interaction. The AI-driven city is sometimes nothing more than the outsourcing of risks and costs to the citizen- users through “platform economies”. These are economic entities that claim sovereignty which previously belonged rather exclusively to the state. Herewith various kinds of intelligent machines operated exclusively by these platforms are powerful instruments to combat for this sovereignty.

There are two people sitting at a McDonald’s in Moscow. One has a Yandex food delivery box waiting on his or her lap. They are having a date, they are kissing. How much time they have left, only the algorithm of the delivery service knows. It will decide when it’s time to break the kiss, buzzing with a new delivery task. Buzzz, the phone goes. One of them darts off and runs away. There are fines for being delayed in the aggregator of the food delivery service Yandex-food.

Nowadays, Moscow is a city filled with people suspended in the waiting mode, dictated by the platform economies and their AI-driven systems. The screens of their smartphones become intimate interfaces – specific social connections between individuals, things, technologies, animals etc., which lead to a unique type of interdependency. On the screen of the smartphone personal and professional lives are mixed. One notification from your lover is followed by a new task of the delivery service that you work for. Thus the private life of contemporary urbanites is being fragmented by moments in which the algorithmically driven economy mobilizes their labor force.

Perplexed Cultural Machines

Breaking the convenient, cozy and familiar scenario of just using the AI-driven tools provided by the platform economies with our project Platform Perplex your next Uber ride is about to collapse. Here, we want to reach these painful points that allow platforms to become not just economic tools but cultural machines born from contemporary algorithmic capitalism.

Intelligent machines that are the heart of uber service work with territories in terms of which there is a need to animate a complex choreography that involves drivers and users. Active work with real time techniques for analyzing big amounts of data based on machine learning gives huge possibilities for predicting and ruling drivers and users’ behavior taking into account the city traffic in general.

Constantly collected data are used for organizing the city as variety of unseen zones with different characteristics. Zones that are changing their shape and size through the time. Visible effects of such a zonation are the moments when a taxi ride is suddenly overpriced or a button “finds trips toward a destination” in an interface of a driver’ application, allowing the driver to get trips on the way home. An example of such hidden effects of the zonation appearing only at the moments of scandals is an intracorporate program related to detection and prevention of “violation of terms of service” activities as it used to be with the attempts to fight the regulators when analysis of the users behavior was used to uncover potential state agents and police officers to block their application.

To start working with cultural effects that produce such algorithmic-based economies we propose to look at this situation as at the landscape where intelligent machines owned by uber perform knowledge production and where human bodies of drivers, passengers and cars are used as manipulators that performs embodiment of the results of that produced knowledge. Having that in mind we are making under a highway a session of emotionality, revealing surrounded computations inside of this platform. Revealing the roughness of algorithms that are given to drivers and users’ outsourcing we work with effects of emotionality that make intelligent machines cultural machines.

Possessing its autonomous agency, acting globally, cultural machines offer a previously unknown level of universality, a universal tool based on symbolic languages. This approach to grasp the present opens up certain possibilities and closes others. The main question that cultural machines put for us is what is knowable and desirable.

Emotions circulate between bodies and interfaces. They define the dramaturgy of relationships that legitimizes those economic relations, relationships of control. They create a velocity, give acceleration to modern disperse economies, adaptable to users. Emotions become commodities along with human attention. And their flows create positions that a contemporary subject can occupy in the soup from the consumer and production activity, where a collision with the intimate knowledge of Siri makes you experience alienated at a radically intimate level.

Invited to the session drivers are caught under siege of the emotional outsourcing. Their cars morph into a laboratory of affects filled with participants to the hilt. We are searching for blind zones where communication acts are dangerous (because of not knowing what can happen), where actors leave their positions, where comfort is destabilized. These zones are temporary – it is inconvenient to stay at this uncertainty, but if you manage to stay longer – then the circumstances that form it are revealed. Hereby we want to shift the focus to the conditions which made the drivers to appear here and to the work that users and drivers need to accomplish for the economy of the uber service to exist.

“De-engineering” of Cognitive Work

Ideas about how the urban landscape of tomorrow will look like is born not in the city square or in the city parliament. City square turns from the public space into a place of gathering of citizens, which can be viewed as a focus group whose task is to give an answer how to improve the economic functioning of the city. And the city parliament is a tool for legalizing corporate decisions taken somewhere outside of traditional places of political city processes.

Political action is transferred to spheres where physical presence is not so important as the presence of tools that disrupt the position of a person as a user that can give it a political dimension that allows challenging the current situation. So it gives the potential to change the political language in such a way that the currently dominant engineering discourse will be politicized. The process that can be called “de-engineering” of cognitive work has the politics of intelligent machines and the politics of production of intelligent machines as the most urgent issues.

Reconfiguration of the bundle of intelligent machines + city where data-sets and a program code are understood as a law, require a reinvention of the political language adequate to the current situation. The question is how to include programming languages in political languages? After all, this means that the repository where the program code is stored is the same public place as the city square where the political process is superseded by the idea of technical progress.

At the same time, we can take the places of these intelligent machines production as a part of broader knowledge production processes. For us the question is — what does it mean to politicize a startup scene as a whole or a haсkathon as a method of producing different types of technological phenomena in particular. Therefore this is exactly the place where knowledge is made about the future of the city. What Does It Mean To Communalize AI Production?

What is specific about the type of knowledge production at a hackathon – a 24 to 72 hours session on collective prototyping of ideas? On the one hand, long-term togetherness forms a dense temporal collectivity. The way of thinking is deformed by this. The extra-subjective consciousness is gaining momentum. It’s like a party that has already culminated and crumbled a few times, but it still lasts. You are locked in a room and no chance to leave. You fall into a dream, but there is no sleep: Everyone around keeps the rhythm of informal production.

On the other hand, it reminds a sports hall, a half-gentrified workshop of an abandoned factory or a lobby of a business center, where such events usually take place. No chance to hide or fence off or to do something by your own. This is a thoroughly visible space of hopes and conflicts — the public space of cognitive workers. What promises us a foray into the hackathon as a form of production? What trophies can we take with us? Or discard as useless? Can we appropriate this form of collective work as a tool to experiment with the specificity of modes of intelligent machines production?

“Hackathon Choreography” is a time-boxed scenario where we accelerated time. 48 hours are squeezed into three hours. 1 day of a hackathon was equal to 90 minutes of our session. 1 hour was equal to 3.75 session minutes. We use acceleration as a method to shift the focus from what is produced on a hackathon to the processes that are standing behind the production of it. Orientations on prospective investors, competition, profitability and investment suitability are the factors that make impact on technologies that are born at a hackathon. This is a place where technological and capitalistic paradigms meet with each other and while interweaving intelligent machines that can survive at the market are born. We however would like to take it a step further and create a situation where one might puzzle out how this form of intensive knowledge production can be useful, deforming the frame in which this production is now written in.

What is an unprofitable AI? How can it exist? What are the preconditions for its development? What does it mean to