For more, check out the previous meeting from the Standing Panel on Technology Leadership here.

The Collision of Exponential Technology and Institutional Legitimacy

We stand at a remarkable moment in history. As the United States marks 250 years of democratic governance, our public and private institutions face a profound operational paradox. Technological capabilities, particularly artificial intelligence, are accelerating at unprecedented velocity. Yet across free societies, public confidence in civic institutions continues to erode at alarming rates.

During a recent National Academy of Public Administration (NAPA) panel entitled “Advancing Trust and Human Agency in AI-Era Governance,” skillfully moderated by Dr. Alan Shark, co-chair of NAPA's Standing Panel on Technology Leadership, leaders gathered to examine this critical intersection. As Dr. Shark observed at the outset, governance has always centered on how people maintain control over their lives, but the arrival of generative and agentic systems forces a fundamental question: " suddenly, we've embraced … our ability to have some control in our lives and everything that we do."

The numbers tell a troubling story. Nearly 70% of citizens express deep skepticism regarding AI, while public institutions are being driven to adopt automated systems to create internal efficiencies. Over 1,100 pieces of AI-related legislation have been introduced across the 50 states in the past two years alone, reflecting widespread legislative anxiety over automated decision-making. Meanwhile, hundreds of billions of dollars in proposed data center projects face mounting community opposition as citizens question who benefits from and who should pay for infrastructure that consumes enormous energy and water resources while delivering opaque algorithmic outputs.

This is not abstract policy friction. It is a legitimacy crisis playing out in real time across local zoning boards, state procurement offices, and federal agency rulemakings.

The Governance Gap: Black Box Decisions and Eroding Mandates

Trust is not an abstract sentiment. Dr. David Bray, Principal and CEO of LeadDoAdapt Ventures, defined trust during the panel as "the willingness to be vulnerable to the actions of an actor you cannot directly control." Humans extend trust when they perceive three essential qualities: benevolence, competence, and integrity. While modern AI models can easily simulate conversational empathy and pass the Turing test, simulation is not integrity. When systems fail silently, erase data, or hallucinate critical outcomes, institutional credibility collapses.

Sharron McPherson, Founder and CEO of Resilience Technologies Inc. and author of From Risk to Resilience, captured the core governance challenge facing local leaders: "The question is: trust from whom, to whom? And who really holds the mandate? We need a common framework for analysis, and scientifically rigorous data that policymakers can use to make decisions that actually fulfill that mandate?"

This tension is playing out daily across state and local governments. Douglas Robinson, Executive Director of the National Association of State Chief Information Officers (NASCIO), pointed out that state leaders find themselves caught between immense pressure to modernize and a turbulent legislative landscape driven by constituent anxiety.

"We talk about trust all the time," Robinson observed. "The citizens' experience with digital government, their citizen digital experience, their CX is if it's positive, then that's a strong predictor of a higher trust level. The interaction that they have, if it's seamless and it's smooth, they tend to trust that government and particularly that agency more because of that experience. Well, what happens when you interject a generative AI tool or an agentic AI agent [when] it could be actually kind of busting that trust equation?"

The answer is already visible. When citizens cannot understand how an algorithm denied their benefits claim, rejected their permit application, or flagged their child for intervention, they do not blame the algorithm. They blame the institution that deployed it. The efficiency gains the government departments celebrate internally become legitimacy losses externally when citizens feel processed rather than served.

Three Strategic Take-Aways for Leaders

1. Mandate Shared Engagement Playbooks and Explicit Decision Rights Before Deployment

Public and private capital investments in automated technologies must be conditioned on structured, transparent frameworks for operational use. Organizations cannot treat AI governance as an ad-hoc experiment conducted by technical teams in isolation from frontline workers and citizens.

Dr. Ramayya Krishnan, Dean Emeritus and Professor of Management Science and Information Systems at Carnegie Mellon University's Heinz College, framed this requirement directly: "No public investment without a credible engagement playbook. So the idea of engagement playbooks, which covers a lot of what we've talked about, which is importantly involving the users in the decision making rather than basically telling them adapt to whatever we're going to create."

The current reality is the opposite. Agencies risk deploying chatbots that might hallucinate legal guidance, procurement systems might auto-reject bids based on opaque scoring, and benefits platforms might deny claims without human review. Each incident could compound public skepticism and legislative backlash. Establishing horizontal playbooks across agencies ensures that institutions do not reinvent basic safety, privacy, and accountability standards with every individual procurement.

2. Shift from Passive Oversight to "Humans at the Helm" Service Delivery

Government agencies and commercial enterprises often make the mistake of deploying automation purely for internal process optimization, neglecting the human impact on the receiving end. Real trust is built when technology directly expands human agency, delivering tangible, visible benefits to the user.

Patricia Cogswell, Partner at Guidehouse and former senior homeland security executive, pointed to the evolution of federal identity and screening systems as proof that public trust follows demonstrated utility: "Start with the mission, design for humans, and make sure you are earning that trust every day." She noted that early biometric travel initiatives initially faced severe skepticism, yet today approximately 80% of the traveling public states that they support use of biometrics because of the speed, predictability, and security benefits.

The contrast with current AI deployments is stark. Citizens encounter automated systems that make their interactions with government slower, more confusing, and less responsive. Call center chatbots that cannot escalate to a human. Permit systems that provide no explanation for denials. Fraud detection algorithms that freeze accounts without recourse. When automation is designed to serve individuals rather than merely process them, citizen engagement transforms from hesitation into active partnership.

3. Build Proactive "Plan B" Resilience and Narrative Transparency

No technological architecture is flawless. In complex digital-physical environments, autonomous agents and algorithms will inevitably encounter edge cases, corrupt inputs, and unexpected failure modes. Resilience depends on pre-engineering fallback mechanisms before deployment rather than scrambling to manage fallout after an incident occurs.

Cogswell outlined the operational discipline required for resilient deployments: "First is you have to actually be able to identify when something's going off the rails. I see too many of these instances, they find out weeks or months after as opposed to as it's happening where they are best positioned to stop it. Number two is proactive risk management... Identify potential risks and implications. Design your mitigations ahead of time, your remediation or what you're going to accept." She added an essential communication rule: "Tell everybody what you're doing and what you're doing about it. And remember, if you don't tell your story, somebody else will."

The current environment is defined by reactive crisis management. Agencies discover algorithmic bias months after deployment. Vendors quietly patch hallucination bugs without disclosure. Procurement offices learn their AI contractor violated data privacy rules through media reports rather than internal monitoring. Each failure feeds the narrative that institutions have lost control of the technologies they deploy or never had control.

McPherson argued that transparency turns contentious public debate into productive governance: when institutions arrive with shared data and a common framework of analysis, stakeholders can move from opposition to negotiation, and from politics to evidence.

Creating the Future of AI-Era Governance Amid AI Concerns

The transformation before us cannot be solved by regulatory prohibition or uncritical technological boosterism. By 2030, more than 40% of the world's digital data will be generated by artificial intelligence - already in 2025, 30% of enterprise content was generated using generative AI tools alone. The volume and velocity of automated interactions will overwhelm traditional, bureaucratic oversight models.

The path forward requires intentional, collaborative statecraft. We must treat data governance, algorithmic accountability, and workforce upskilling not as compliance burdens, but as foundational public infrastructure.

As Dr. Shark observed in his closing reflections, machines will never possess the uniquely human capacities for moral deliberation, genuine empathy, and open-minded curiosity. Our responsibility as leaders across the public, private, and civic sectors is to ensure that technological tools remain anchored to human purpose.

As Dr. Bray concluded: "The best way to predict the future is to create it. So we need to lean forward with things that actually are projects, infrastructure, are involving people, doing the business of governance with people as opposed to to people in a new model forward." By establishing robust playbooks, placing humans firmly at the helm, and fostering radical transparency across our institutions, we can build an AI-era governance model that expands freedom, strengthens human agency, and secures public trust for generations to come.