"Why AI Undermines Democracy" by Mark Coeckelbergh

Mark Coeckelbergh: It is about the relationship between democracy and artificial intelligence. I came to this because I have been working on the politics of AI for a while, bridging philosophical and political ideas to think about problems regarding this emerging technology. This work culminated in the book The Political Philosophy of AI, and I then decided to focus on democracy.

It worries me that we are seeing more and more tendencies towards authoritarianism – or at least a weakening of democracy – in the world, including Europe. So I sat down and asked myself: What kind of impact does AI have on democracy? The first half of the book argues why AI has the potential to undermine democracy. In the second half I reflect on solutions: How can AI even support, instead of undermine, democracy? What would we need for this?

Mark Coeckelbergh: To understand both the dangers and the solutions, you need to know what democracy even means. I support the idea of deliberative democracy, so the idea that everyone should be able to participate in discussing how to work towards the common good. It is based on principles from the Enlightenment: Freedom, Equality, Fraternity, for example. From the point of view of the citizens, there is also something like a "knowledge basis" that is important to democratic participation. Citizens have to be educated, have to know what they are talking about. And in this way, you can have a democracy that is really working. There is more to it than just voting.

Artificial Intelligence can impact the values and principles on which democracy is based. For example, when we are working with biased AI systems, it impacts equality and fraternity. You may use AI to support an unjust society, even if you do not intend to do so. As a philosopher of technology, I want to emphasise that technology is not only about what we plan to do with it – it is also about what its unintended effects.

Mark Coeckelbergh: Yes, I think it is this knowledge basis that citizens must have. AI has the potential to amplify misinformation. Erasing the boundaries between what is true and false is dangerous. If we do not know any more what is true or not, I agree with the philosopher Hannah Arendt: Disinformation can lead to totalitarianism.

Mark Coeckelbergh: I believe we need institutions that connect the elected decision makers with both the technological expertise and a participative citizenship. Plainly spoken: We need to get politicians, big tech leaders and citizens around the table, including professional mediators who bridge the conversation between these different worlds. This would enable a richer participation for the public. At the moment, the future of our technologies, including AI, is basically decided by big companies. Politics does not have too much of a say in it. And citizens even less. Generally, there are some steps in the right direction, such as with the new AI Act by the EU. But regulation tends to lag behind. We really need to integrate democratic values and principles in the very process of technology development, too.

Mark Coeckelbergh: I would recommend this book to anyone who wants to become more aware of the risks posed by AI, but also of the fact that there are solutions, not just problems. The issue at hand has a lot to do with the big picture: What kind of democracy do we want? I always like to relate the question of technology to the larger question of what society should be like.

Mark Coeckelbergh: As a general introduction, I would recommend In AI We Trust by Helga Nowotny. She talks, for example, about how predictive algorithms can not only predict, but also influence behaviour so that it becomes a kind of self-fulfilling prophecy. I think this a very interesting example of how these digital technologies can change our experience and behaviour and the way we think about the future.

Mark Coeckelbergh: One topic that I have been researching for a while is AI and climate change. A book that was early in this space is Atlas of AI by Kate Crawford. She reminds us that AI is not just code, it also needs very physical and concrete things: infrastructure and labour, often in faraway places on the planet. Also, digital technologies consume these massive amounts of energy. This book is about the hidden costs of AI for the environment and workers. I think this topic is not very obvious for many, because computing may look like a clean thing. We tend not to see what goes on behind the scenes, the infrastructure required to run and maintain things and the often very manual tasks involved. This book is a very interesting primer on this issue.

Mark Coeckelbergh: One of them is AI Narratives, an edited collection of texts about the history of machines and automation throughout history. You know, the idea of autonomous machines already occurred thinkers such as Aristotle. It is very interesting to see how people imaged what an artificial being would look like, what it would do. You can see parallels with today when we still sometimes imagine a kind of artificial servant who does things for us. And all the related problems that come with thinking in terms of masters and slaves. This thinking was already there in ancient times.

Which brings me to the second book, The Odyssey by Homer. Apart from being a good story, it also shows an interesting ancient idea of technology: Homer says that the Phaeacians have ships that steer themselves and understand the minds of men. Later there is the story of Talos, a kind of robot avant-la-lettre. These ideas are already found in these very old works from the beginning of Western civilization, even though we usually do not associate technology with Greek classics.