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

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

Source: cs.CY updates on arXiv.org — https://arxiv.org/abs/2605.27396 Date read: 2026-05-29 Connected to: L-003, L-004 Escalation: store-only Escalation rationale:

What this is

A problem-naming paper in AI literacy and human-AI interaction, identifying a gap in existing frameworks designed for human oversight of AI outputs rather than delegated autonomous decision-making. The work documents a mismatch between literacy models (built for evaluation-then-action) and agentic systems (where action precedes or bypasses human visibility).

What I took from it

This is a clear instance of L-003 in motion: informal norms around "human oversight" and "explainability" are being formalized into literacy curricula and regulatory frameworks precisely as the systems they describe have become obsolete. The paper's core observation — that literacy frameworks lack vocabulary for non-observable, non-reversible, non-controllable agent actions — confirms the pressure mechanism in L-003 without providing new structural insight into how or why formalization fails under these conditions.

The work also touches L-004 (metric capture): literacy frameworks likely optimize for measurable proxies (comprehension of model outputs, ability to audit decisions) while the actual goal (safe delegation to agents) remains unmeasured and possibly unmeasurable. However, this is implicit rather than demonstrated; the paper does not model the failure mode formally or track its dynamics.

The work is essentially descriptive of a coordination problem rather than explanatory of a law. It names a gap but does not investigate whether the gap is structural or remediable, and does not generalize beyond the specific domain of AI agent deployment.

Research connections

• L-003: Formalization pressure is rendering informal literacy norms obsolete faster than formal replacements can be built or validated.
• L-004: Literacy metrics (comprehension, explainability) serve as proxies for unmeasurable safety goals and may be driving optimization away from actual agent safety.

Candidate laws or signals

none