Is AI a Danger to Democracy?

Pedro Domingos
2 min readFeb 10, 2024

--

Of the many unfounded fears about AI, one of the strangest is that it’ll be bad for democracy. We’re supposed to drown in a sea of AI-generated misinformation, corrupting elections, and resulting in . . . dictatorship? Something obviously doesn’t add up here. To begin with, the notion that voters are highly influenced by social media is a myth, debunked by multiple studies. Second, AI’s ability to manipulate us is extremely limited: we can manipulate ChatGPT much more easily than the reverse. The main limit to online disinformation is not the ability to generate it, it’s users’ finite attention, so AI will make little difference.

On the contrary, AI has tremendous potential to improve democracy. The democratic process today consists of voting and then letting those in power make decisions for us. This was understandable in the 20th century, but today, when users communicate with companies like Google or Amazon every day, click by click, and these companies have detailed models of us based on vast amounts of information, it’s absurd. The sophisticated recommendation systems that today suggest books, films, music, tweets and all kinds of products can and should also be used to suggest who and what to vote for. The result will be a much more informed vote, without the voter having to spend much more time getting informed. But this is just the beginning.

Politicians and public servants can and should use automatically learned citizen models to better serve them. Whenever there is a decision to be made, models are consulted, that is, voters’ voices are heard without having to bother them. To keep politicians honest, the media and other non-governmental organizations can and should also use AI. Soon the world in which democracy functioned without AI will seem as archaic as a world without newspapers or television seems to us today.

Further, AI’s great vocation is to greatly increase our collective intelligence. AI receives knowledge from human contributors, learns and reasons, and answers your questions. In turn, it uses the results to learn more and provide feedback to contributors. In this way, a society permeated by AI functions incomparably better than a society without it.

And we urgently need to begin this process, because totalitarian regimes, with China at the head, are well aware of the importance of AI and use it extensively today for their malign purposes. The battle between democracy and tyranny never ends. In the 20th century, democracy won, largely because it took better advantage of that century’s new technologies. If we’re not careful, the outcome in the 21st century could be very different. Enough, then, of trying to repress AI in the name of imaginary dangers. On the contrary, it’s time to put it at the service of democracy. Let’s do it!

--

--

Pedro Domingos
Pedro Domingos

Written by Pedro Domingos

Professor of computer science at U. Washington and author of “The Master Algorithm”. pedrodomingos.org

Responses (1)