Belgian philosopher Mark Coeckelbergh (*), specialised in media and technologies, is one of the 40 experts named by the United Nations to form an Independent International Scientific Panel on Artificial Intelligence (AI).
The experts were selected from more than 2,600 candidates after an independent review by several UN bodies, including the International Telecommunications Union (ITU). They will serve for a three-year term, to advance scientific understanding and ensure that international deliberations on AI are informed by the best available evidence.
Their annual report will feed discussions within the Global Digital Compact adopted at the 2024 Summit of the Future, that set out a shared vision for an open, safe, and inclusive digital future.
“In a world where AI is racing ahead, this panel will provide what’s been missing — rigorous, independent scientific insights that enable all Member States, regardless of their technological capacity, to engage on an equal footing,” said UN Secretary General António Guterres.
Mark Coeckelbergh underlines the importance of global governance for AI in times when multilateralism is under attack.
You have written several books on AI, notably AI Ethics, and Why AI Undermines Democracy and What To Do About It. In the latest book, you demonstrate that the current trajectory of AI is bad for humanity. Can you tell us why?
As it’s currently used and developed, AI undermines democracy in a number of ways. It can be used to manipulate voters and elections in a manner that was unseen before. What is also new is the personal targeting of voters based on their personal data.
More worryingly, in my view, is that the basis of our democracies is undermined. The principle of equality is, for example, undermined by huge power asymmetries between individual citizens and Big Tech. The principle of freedom is compromised by the surveillance tools that are now available.
The kind of information environment that is created is also problematic. We can no longer be certain of what is true and what is not. The philosopher Hannah Arendt already warned us that anyone wanting to control citizens is served by confusion. Propaganda then works better, and it could lead to totalitarianism.
In the second part of the book, I show how we can change things and change the environment in which AI is developed and use it for democracy.
You co-authored a book called “Communicative AI”, what is the impact of this technology?
It is about large language models (LLMs) and their philosophical significance. Suddenly, we have a tool that can also produce text, something that previously only humans could do. Because we are not machines, and not like other animals.
There is a philosophical tradition where we are defined through words. Now, machines produce words. In this book, we reflect on what it means for the notion of authorship, and for language and communication. When we write with LLMs, for example, it is clear that we are no longer authors in any traditional sense of the word. But what are we then? And is it still writing? We are forced to re-think what it means to be an author and to write.
AI threats range from misinformation to environmental harm, and perhaps even AI getting out of control… What is the biggest risk in your view?
I focused on the political risks, but there is also a big risk that’s more psychological, with the addiction phenomenon and our dependency on technology.
The risk is not so much about AI becoming conscious or taking over, but rather very concrete risks for education and the exposure of vulnerable people, such as children, people with mental health problems or people who are very lonely. These problems will become more pronounced as chatbots are increasingly convincing in having the appearance of a person.
How, in your view, is AI changing our perception of truth?
With AI it’s easy to produce misinformation. This creates a situation in which it’s no longer clear what is true. Any information can be false. Any quote, video, or sound recording can be fake. This confusion is dangerous for democracy.
In a context where global governance of AI is lacking, will the panel be able to make recommendations that could become binding?
Very good question. The work of the panel and advisory bodies is very meaningful, even in a context where multilateralism is under pressure. In my personal work, I argued in favor of global governance of AI, to set guidelines and provide directions. Even if it’s unlikely that binding rules will be adopted any time soon, it’s important to prepare for this.
The panel is seen as scientific, above all, so the emphasis here will be less on governance and more on giving scientific expertise. But of course I will report on the expertise in AI ethics and that includes all these concerns.
Do you use social networks yourself?
Yes, of course. As an academic nowadays, you have to let the world know that you exist and publish your results. And also privately, I use social media to keep in touch with people.
What is a recent example of disinformation that struck you?
What I find striking is that, for example, on platform X, formerly Twitter, Elon Musk really seems to push some narratives rather than others and suppresses opinions he disagrees with. I really see that this can be a very effective means of propaganda in the current times. That’s one of the reasons why I’ve written an article calling it “techno-fascism”.
Is there anything that gives you hope for the AI-driven future?
Yes, I see that more people today see the dangers and challenges associated with AI. I hope this will ultimately translate in guiding the technology in more ethical and democratic directions. AI should be a technology that benefits us all.
Mark Coeckelbergh, Professor of Philosophy of Media and Technology at the University of Vienna, holds a Chair at the Institute of Philosophy of the Czech Academy of Sciences, where he leads the Center of Environmental and Technology Ethics. He regularly advises policymakers and international organisations, including the European Commission (High-Level Expert Group on AI), the Austrian UNESCO Commission, the Austrian Council on Robotics and AI, and the Belgian Federal Advisory Committee on Data and AI.