Presented by EY
In our previous episodes, we explored steady growth and explosive transformation. But what happens when an algorithm makes a mistake, deep fakes flood the zone, and you can't find a human to fix it? In the third episode of The Four Futures of AI, we enter the "Constraint" scenario. Hosted by Francis Rose, this episode explores a future where the blinding speed of AI innovation hits the brick wall of public opinion, forcing governments to slam on the brakes to prioritize human control over raw speed.
Rather than letting technology run unchecked, the "Constraint" future envisions a world where public backlash from a major AI failure forces society to demand strict regulation, accountability, and AI assurance. Experts in this episode discuss how trust is a strategic asset that takes decades to earn but can be destroyed in an instant, and how to safely navigate the deployment of AI across four key domains:
The Trust Deficit
The episode begins by exploring the growing gap in public trust. As AI is increasingly thrust upon consumers and citizens, the potential for catastrophic mistakes, data privacy violations, and sophisticated AI hackers subverting systems creates a volatile environment. Experts compare this potential future to the history of nuclear energy—where high-profile disasters like Chernobyl caused massive public backlash and stalled innovation for decades. To avoid this, governments must proactively focus on security and building systems that are actually trustworthy.
Policy, Procurement & AI Assurance
How does a deliberately slow and steady legislative process keep up with technology moving at breakneck speed? This chapter dives into the scramble to build proper AI governance and procurement strategies. Experts discuss the critical need for "AI Assurance"—essentially auditing models to see how safe they are, why they make certain decisions, and identifying their biases. The conversation highlights the need for taking responsible risks, including a proposal to treat AI failures like airplane crashes by using an NTSB-style investigative body to learn from mistakes and prevent them from happening again.
The "Human-In-The-Loop"
While AI can rapidly process information, it cannot replace human instinct or accountability. This chapter focuses heavily on the imperative of keeping a "human-in-the-loop." Experts and state IT leaders draw a hard line: AI should never autonomously make life-altering determinations, such as denying welfare benefits, deciding who gets priority services, or initiating law enforcement actions. The episode reinforces that while AI is a powerful tool for efficiency, the ultimate responsibility to the citizen must always remain with a human being.
The Global Balancing Act
The episode concludes by addressing the massive geopolitical risk of the Constraint scenario: if the U.S. locks down innovation with heavy regulations, do we surrender our advantage to adversaries like China and Russia? Experts grapple with how to balance the need for safe, responsible AI development with the necessity of winning the global AI race. Ultimately, the chapter reframes constraints not as a negative restriction, but as a safe boundary—providing the guardrails necessary to give the government workforce the confidence to experiment and innovate without the fear of making a mistake.
--------------------------------------------------------------------------------
Appearing in this Episode:
-
Francis Rose, Host, Fed Gov Today
-
Jim Bryson, Co-Chair, AI Advisory Council, Tennessee Department of Finance and Administration
-
Dan Cronin, Chief Information Officer, State of Oklahoma
-
Ken Dircks, Infrastructure & Transportation Solution Lead, EY
-
Ryan Doherty, Principal, Government & Public Sector Technology Consulting, EY
-
Dennis Eger, Senior OSINT Advisor, Army
-
Jennifer Huddleston, Senior Fellow in Technology Policy, Cato Institute
-
Amy Jones, U.S. Public Sector AI Lead, EY
-
Dave Larrimore, Former CTO and CAIO, DHS
-
Frank Lee, Defense Sector Lead, EY
-
Dawn Meyerriecks, Former Deputy Director for Science and Technology, CIA
-
Brian Peretti, Former CTO and CAIO, Treasury Department
-
Dru Rai, CIO, State of New York
-
Mark Roberts, Manager, Artificial Intelligence Engineering, EY
-
Nicol Turner Lee, Director, Center for Technology Innovation, Brookings Institution
-
Bruce Schneier, Lecturer in Public Policy, Harvard Kennedy School
-
Nishant Shah, Senior Advisor for Responsible AI, State of Maryland
-
Alan Shark, Associate Professor, Schar School for Policy and Government at GMU
-
Rep. James Walkinshaw, Congressman (D), Virginia
-
Zach Whitman, CDO and CAIO, GSA
-
Rep. Rob Wittman, Congressman (R), Virginia