Agentic Literacy Debt: A Structural Problem the AI Literacy Field Has Not Yet Named

The current AI literacy framework is obsolete for autonomous agents: Existing AI education was designed for scenarios where humans review AI outputs before deciding to act. However, autonomous agents now make decisions and take actions on behalf of users in healthcare, finance, and workplaces without step-by-step human approval, creating a gap that traditional literacy training cannot address.

The costs of this "literacy debt" fall on users, not deployers: Organizations that deploy autonomous agents incur the debt, but the consequences—unpredictable or irreversible actions taken without user visibility or control—are experienced by patients, citizens, and end users who have delegated decision-making authority to systems they may not understand.

The problem is structural and self-reinforcing, not temporary: The paper identifies three reinforcing mechanisms that compound the problem: normalization of opaque delegation (users accepting they can't see what agents do), multi-agent ecosystem complexity (multiple agents interacting in unpredictable ways), and institutional path dependence (organizations locked into agent-based systems). This suggests curriculum reforms alone cannot solve it.

Real-world harms are already occurring: Evidence from healthcare, financial fraud detection, and global equity contexts demonstrates the gap is "already consequential"—not a theoretical future problem—indicating autonomous agents are causing tangible harm today without adequate user understanding or safeguards.

AI literacy must become a governance issue, not just an educational one: The paper argues that addressing agentic literacy debt requires reframing AI literacy as a governance capability, suggesting the solution involves policy, institutional design, and system accountability—not just teaching people to better evaluate AI.