Democratizing AI
Calls for ‘democratizing AI’ are everywhere, particularly during the latest generative AI deployment wave. While this wave has prompted escalating fears about AI posing an existential threat to humanity, there has also been an uptick in unbridled techno-optimism about the positive potential of a large number of people starting to integrate generative AI tools into every aspect of their daily lives. Surely, the thinking goes, this will ‘democratize AI’ by making sophisticated technology much more accessible to ordinary citizens without specialized expertise.
But it would be misguided to jump to that conclusion.
As a democratic constituency, we currently find ourselves in a choice environment marred by an asymmetrical concentration of power, wealth, and information in the hands of a small group of AI deployers, coupled with a widespread dispersion of AI-related risks across the demos at large. In this book, I argue that this asymmetry creates a particular type of democratic legitimacy deficit that warrants a particular type of collective response.
AI deployers, most of whom are oligopolistic corporate actors partial to fundamentally antidemocratic political ideologies and not subject to direct democratic authorization, are currently able to set AI regulation agendas unliterally. They do so either by increasingly intervening in governmental institutions directly, or by the mere act of simply developing new tools and then granting public access to them. At the same time, these actors offload responsibilities for dealing with AI deployment’s potential harmful consequences onto government actors and—by extension—democratic constituencies.
just because ai is at all of our fingertips, that does not mean it’s truly in our hands.
the time for meaningful democratic control over ai’S FUTURE DIRECTION is now.
You can pre-order the book. Democratizing AI, forthcoming summer 2026, now.
Journalists & book critics may request an embargoed advance reader copy here.
→ More about the author, political philosophy professor Annette Zimmermann
→ Featuring critical commentary by Thomas Christiano & Ritwik Agrawal, Chiara Cordelli, Regina Kreide, Mathias Risse, Allison Stanger, and Nadia Urbinati.
Once democratic constituencies relinquish the power to effectively set AI policy agendas, and find themselves locked in a dynamic of reacting to an unduly narrow and risky cluster of options presented to them by an unrepresentative subset of the democratic constituency, they forfeit an essential feature of democracy itself: the right to exercise free and equal anticipatory control over the nature and scope of decisions shaping their shared future.
Democratic constituencies need to stop uncritically accepting AI exceptionalism: the view that AI is an inherently mysterious, highly unusual policy domain that cannot possibly be subjected to democratic scrutiny.
Instead, all of us—as democratic constituencies—must identify ways of taking charge, inspired by the ways in which other complex science and technology policy domains have been democratized. This book offers an AI democratization Playbook outlining concrete strategies for taking back control—by harnessing the power of tech workers, by amplifying the democratic impact of independent expert voices in AI policy, by identifying innovative ways of mobilizing ‘regular’ citizens, and by rethinking structures of ownership in the technology sector.
current Talks & Events
APRIL 3: COLLOQUIUM IN LAW, PHILOSOPHY, AND POLITICAL THEORY, UC BERKELEY, OAKLAND, CA, USA.
APRIL 8: POLITICAL THEORY COLLOQUIUM, YALE UNIVERSITY, NEW HAVEN, CT, USA.
MAY 8: WORKSHOP AT LEIDEN UNIVERSITY, DEN HAAG, NL.
MAY 18: RE:PUBLICA. BERLIN, GERMANY.
JUNE 9: WORKSHOP AT COLUMBIA UNIVERSITY, PARIS CAMPUS, PARIS, FRANCE.
JUNE 11: WEIZENBAUM CONFERENCE, BERLIN, GERMANY.