Democracy is not a static thing. It is constantly being made and reshaped by every possible force. In the twentieth century, the idea of self-governance by the people was threatened by new shifts in media, as radio and then television emerged, followed by the early internet at the end of the millennium. These shifts in technology happened alongside war, disease and economic unsettlement, leaving people trying to make sense of a society moving faster than it ever had before.
AI has arrived at another such moment, bringing with it a technological revolution unlike anything we have seen before. It is a fundamentally different kind of technology, one that can operate outside of direct human control and using levers that humans can no longer understand or see. If politics is the art of making things possible, AI, through algorithms and generative forms, has added a kind of unknowable alchemy to that process.
As with past shifts, this has required a new generation of scholars, writers, critics, and advocates to emerge as society grapples with this question of the impact of AI on democracy. Among them is Northeastern University’s Beth Simone Noveck, who is also chief AI strategist for the US state of New Jersey.
Noveck’s Reboot is a serious book about a real problem. It is also, in important ways, the wrong book for this moment. It looks at AI as an institutional problem of democracy, not a framing I had encountered in any meaningful way before, and certainly not from someone with the experience and knowledge to speak with real authority.
The book is essentially a cross-stitch of case studies showing how AI can be implemented in the public sector to improve service delivery and help citizens engage more meaningfully with a government that has grown monstrously in its remit but shrunk in its capacity to give citizens a voice in their own governance.
Noveck’s work is at its best when it treats democracy not as an abstract concept, the thing we all love to eulogise, but as a set of administrative problems that can be improved with the thoughtful application of tools already revolutionising the private sector and so much of our daily lives.
That premise is also what makes the book frustrating. Noveck writes about democracy with the assumption that its crisis is one of institutional capacity and insufficient information processing. There are of course many places where AI can help, but the book treats these as process fixes devoid of cultural meaning, beyond the political will to implement new technologies in a system with planetary inertia.
Which is not to say the book is naive. Noveck makes sharp points about institutions as information-processing systems, and about how new AI tools can help manage the glut of data the modern world produces. The opportunities for better service delivery and for bridging gaps between citizens are real, and the case studies she mentions were the talk of the Smart Cities initiatives in Australia and elsewhere in the late 2010s.
But that was an era of optimism around these technologies, a very different world from the one we inhabit now, something Noveck only passingly addresses. The core tension I found on reading Reboot is that each of its case studies presumes a society comfortable with a new, AI-powered government. It is interesting and worthwhile to discuss how services might be improved, but it is lacking, irresponsible at worst, to give only short shrift to the citizens over whom this new AI-powered government would have authority.
On a stylistic note, this is a book very much written for the AI age. It repeats ideas and characters with the consistency of a television show designed to be watched in the background, making it perhaps the first book I have read made for the second screen. Noveck also writes consistently in the first-person plural. Everything is we. Whether that reflects academic inclusivity or an acknowledgement that we are never quite alone anymore, even reading on a plane, I could not honestly say.
Reboot feels, at times, like an extension of the American liberal obsession with fact-checking in the Trump era. Both share the same foundational assumption: that the dysfunction of democratic life is primarily a supply problem. Give people better facts, or give institutions better data, and the system self-corrects. But democratic dysfunction today is not a supply problem. It is a crisis of trust, and trust is not fixed by better processing. Lies are more powerful than truth because they tell a better story. Grievance is more honest than facts to citizens who sense the modern world collapsing under their feet.
Noveck’s techno-optimism is perhaps best encapsulated toward the end of the book, when she sketches out a future for representative democracy in which AI avatars of both voters and representatives debate the direction of the country, leaving the humans themselves alone. That is a dystopian vision beyond most science fiction. Even with the caveat that humans could still take back the wheel, it would describe a world in which the role of citizen is reduced to that of a driver half asleep at the wheel.
Noveck also gives the example of urban planners in Brazil and Indonesia tasked with deciding the future of their cities’ public transit systems, and suggests AI might have made better decisions than they did. That might well be literally true, but it gets at the core problem of the book. Thirty years of public–private optimisation has left public services so hollowed out that citizens no longer feel they live in a democracy worthy of the name. Democratic legitimacy is not a byproduct of good outcomes. It derives from participation, from citizens feeling that decisions were made with them rather than for them. A better bus route decided by an algorithm is still a bus route nobody asked for. To fight one brand of optimisation with another is to entirely miss why citizens are so frustrated right now.
There is a sadness that comes through in Reboot around the fact that the United States lags so many other countries in implementing technology in governance. Noveck laments that the home of Silicon Valley has not yet benefited from many of these adaptations, but she never quite answers why.
The answer may lie in the fact that these ideas belong to a time when technology was seen as something that could constrain the overreach of government. Now, much of the public views tech companies themselves as the overreach that needs constraining.
Noveck spends much of the book on city-level implementations, where many of her ideas make genuine sense. Municipalities are not tarred with the same brush as national governments, but they exist within the same political context.
These are, in the end, two versions of the same problem. Noveck’s techno-optimism is not just politically naive. It leads her to write a book about public administration when she thinks she is writing a book about democracy. The scope she chose reflects the assumptions she made.
Reboot is less a book about improving democracy in the age of AI than a book about how public sector services can be improved through the thoughtful application of new technologies, an argument that the private sector should not be the sole beneficiary of such innovation.
That makes it a worthwhile read for city administrators and those interested in the finer points of public administration. But anyone hoping to understand the relationship between the rising tide of AI and the very real concerns we should all have about our democracy will be left wanting. •
Reboot: AI and the Race to Save Democracy
Beth Noveck | Yale University Press | $67.99 | 384 pages
