Mark Coeckelbergh is Professor of Philosophy of Media and technology at the University of Vienna, as well as the author of various titles on ethics, technology and politics. In The Political Philosophy of AI, he draws on his extensive knowledge of relevant research and strands of thought to offer an introduction to political thinking in relation to the field of AI. The work is in part grounded in a challenge, stemming from the philosophy of technology, to the naïve notion that technology is neutral and that how we use it is what raises ethical concerns, but is also intended to demonstrate that there a rich body of thought in political philosophy (and not only ethics) that can be applied to developments in artificial intelligence. Above all, however, the book proceeds from the premise that AI is already inherently political. Indeed, as the author writes: ‘… AI does not just function as a politically neutral tool in the hands of humans playing a particular political game, but transforms the conditions under which politics is done’ (page 83). So important has technology become, the he claims, ‘… political philosophy in the 21st century can no longer be done, and should no longer be done, without responding to the question of technology’ (page 150). Thus, the book is built on the position that ‘… the issues we currently care about in political and societal discussions, such as freedom, racism, justice, power, and (threats to) democracy, take on a new urgency and meaning in the light of technological developments such as AI and robotics, and that political philosophy can help to conceptualize and discuss these issues and meanings’ (page 149).

Each chapter considers certain core concepts in relation to AI, including freedom, equality and justice, democracy, power and non-humans. Within each chapter, the fundamental concepts to be discussed and applied are explored, and central issues surrounding AI are examined. For instance, the chapter on democracy deals not only with different understandings of democracy, but also with questions of manipulation, misinformation, populism, totalitarianism and the emergence of echo chambers, amongst others. The key concepts are then brought to bear on the issues outlined at the start of each chapter, the author discussing the ways in which the nature, use and development of AI raises issues in relation to these concepts, as well as the manner in AI can require us to re-think the concepts themselves. The book is, therefore, not just a work of applied political philosophy, but also one of political-philosophical thinking itself. As Coeckelbergh mentions, there is a good deal of overlap between chapters, which means that questions raised in one chapter could just as easily have been placed within another and are sometimes re-visited. As such, the breadth of coverage is impressive.

While the individual chapters address various questions, certain themes or concerns appear constant. These include:

– data collection and surveillance
– the manipulation of choice or ‘nudging’
– the shaping and (re-)constitution of the self through engagement with AI
– censorship and the filtering of information or opinion
– the growth of (unaccountable) power and non-transparent decision making
– errors in algorithms and the reproduction or propagation of bias, discrimination or domination
– issues of exploitation and autonomy

These themes are of course inter-related: collecting data about users makes manipulation possible, which naturally raises questions about personal freedom. As the book proceeds to address these issues using different concepts from political philosophy, there is an element of repetition, but this is perhaps to be expected – and depending on their own concerns and interests readers are likely to find some chapters more engaging than others. For instance, in the chapter on democracy, the author draws on the thought of Hannah Arendt (pages 89-91) to consider the possibility that AI systems, in their presentation and manipulation of information and opinion, risk bringing about conditions from which authoritarianism and totalitarianism might emerge, but also asks whether AI can also help to produce the conditions that enable democracy to flourish. The chapter on non-humans raises interesting questions about whether the effects of AI on animals and the environment should be considered, as well as whether AI systems themselves might come to form part of the political community, and if so, what this might entail. In both cases, a complete transformation of our understanding of ‘the political’ would be required, and we would need to ask ourselves about the criteria by which we might decide to accord non-human entities political recognition: whether the apparent presence of certain characteristics such as sentience or consciousness, the ability to perform certain tasks, the development of certain interdependent relations with human beings, or perhaps some notion of intrinsic or fundamental value.

The conclusion calls for more research on AI, as well as engagement and discussion on the part of various stakeholders throughout society. In an increasingly globalised world in which AI is not constrained by the borders of nation states, the author urges us to avoid reproducing the presuppositions of anglophone political philosophy in our deliberations, particularly if there is to be international co-operation and perhaps a need for supranational institutions and forms of governance to address the challenges presented.

The book is full of interesting questions and considerations from political philosophy and seeks to provide ‘some substantial building blocks for an evaluative, normative framework for thinking about the political aspects of AI’ (page 150).’ However, the analyses and discussions move at considerable speed, as though the author was constrained by a tight word limit. The text moves from one issue to the next very quickly, without sustained consideration of the issues at stake or the full implications of what is under discussion. In the chapter on freedom, for instance, Coeckelbergh mentions Hegel’s dialectic of master and slave, as found in his Phenomenology of Spirit. A somewhat complex idea, particularly for one new to political philosophy, this is covered in a single sentence, before moving to discuss Marx’s thought in relation to freedom and technology. More than anything else, the book would have benefited from a greater volume of concrete, illustrative examples. These would have helped to maintain the more measured pace that one would expect from a book presented as an introduction, spelling out the issues in play before moving to the next consideration. Indeed, the book opens with a very good example of AI failure and its implications, but too often in what follows, instead of a developed example, an in-text reference is given for a relevant piece of work. This creates the impression that the book is aimed primarily at scholars seeking an overview that provides direction for more advanced research on particular issues. The Political Philosophy of AI certainly enables reflection on why certain issues are raised in connection with AI matter and what principles are at stake. It thus helps us to understand the political nature of the debates surrounding AI, and how they are centred on the interplay – and differing conceptions – of certain fundamental concepts and values. Those new to either philosophy or the issues raised by AI will, however, be better served by a more genuinely introductory treatment.

‘The Political Philosophy of AI: An Introduction’ by Mark Coeckelbergh was published in 2022 by Polity (ISBN: 978-1-5095-4854-5). 186pp.

 

Neil Jordan is Senior Editor at the Centre for Enterprise, Markets and Ethics. For more information about Neil please click here.