BOSTON, January 15, 2026 – A new paper from Boston University law professors warns that the very design of artificial intelligence systems poses an existential threat to core democratic institutions, from the courts to schools and even families. The research suggests AI isn’t simply a tool, but a force actively dismantling the structures that underpin modern society.
AI’s Destructive Design: A Looming Threat to Institutions
Table of Contents
The study argues that current AI systems, by their inherent nature, weaken civic institutions to the point of collapse.
- The paper contends that AI’s design is fundamentally at odds with the cooperation, clarity, and accountability vital for healthy institutions.
- Authors Woodrow Hartzog and Jessica Silbey initially intended to write a more optimistic piece, but their research led them to a far more dire conclusion.
- The critique focuses on “predictive and generative AI,” the current wave of AI technology, rather than hypothetical future iterations.
The core argument, laid out in the preprint paper How AI Destroys Institutions, is that AI isn’t a neutral efficiency tool. Instead, it actively degrades the functions of essential civic bodies like administrative agencies and universities. The authors assert that AI systems are, in effect, a “death sentence for civic institutions.”
A Shift in Perspective
Hartzog and Silbey’s research began as a follow-up to a 2018 paper, “the Upside of Deep Fakes,” which posited that deepfakes merely exacerbate existing vulnerabilities
The illusion of Neutrality
The authors directly challenge the notion that AI is merely a neutral tool. “AI boosters defend the technology’s role in dismantling our vital support structures by claiming that AI systems are just efficiency ‘tools’ without substantive significance,” they write. However, Hartzog and Silbey argue that the very architecture of these systems undermines the core principles of institutions, such as administrative agencies and universities.
This isn’t a critique of all possible AI, but a specific warning about the dangers of current “predictive and generative AI” technologies.The paper urges a critical examination of how these systems are designed and deployed,before the damage becomes irreversible.
Worth a look
