Position: AI Lock-In Is in Progress, and We Must Be Prepared

TEXT START: AI safety research has mainly focused on two areas: technical alignment (ensuring AI systems produce human-aligned outputs) and the regulation of generative AI's societal impacts (including unemployment risk and labor market disruption).

The Dissection

The paper identifies AI dependence as a safety problem: automation erodes human skills, making individuals and institutions brittle when systems fail, are attacked, or become inaccessible. Its real function is to convert the civilizational consequences of AI adoption into a manageable resilience agenda—maintain backups, preserve skills, and prepare for outages.

Under the Discontinuity Thesis, this is a genuine secondary symptom, not the primary rupture. The paper studies what happens when the machine disappears. It largely avoids what happens when the machine remains available and makes human labor economically unnecessary.

The Core Fallacy

It treats AI lock-in as excessive reliance that can be voluntarily mitigated. Under P1–P3, dependence is not primarily a choice; it is the competitive result of AI becoming cheaper and more capable than human labor. Human fallback capacity then becomes an expensive subsidy that institutions cannot preserve at scale.

The paper also conflates operational autonomy with economic autonomy. A person may retain the ability to function without AI while losing ownership of the productive systems that determine income, access, and status. Skill retention does not prevent the mass employment-to-wage-to-consumption circuit from being severed.

Hidden Assumptions

• Human-only capabilities can be maintained without becoming competitively obsolete.
• Governments can coordinate durable safeguards against firms and states optimizing for AI deployment.
• AI dependence is reversible before it becomes structurally mandatory.
• National security is mainly threatened by AI outages, rather than by control of AI capital consolidating in Sovereign hands.
• Preserving human competence preserves meaningful economic participation.

Social Function

Partial truth functioning as transition management and ideological anesthetic. The paper correctly describes deskilling and systemic fragility, but packages the transition as a preparedness problem instead of a power-transfer problem. Its recommendations can preserve emergency competence during disruptions; they cannot restore broad productive necessity once AI dominates cognitive work.

The Verdict

The paper diagnoses the scar and mistakes it for the wound. AI lock-in is not merely humans forgetting how to operate without machines; it is the institutionalization of machines as the operating layer of society while ownership and productive authority migrate to Sovereigns. Its mitigation program may reduce outage damage, but it does not challenge the Discontinuity Thesis. It is a resilience memo for the interregnum, not a mechanism for preserving the post-WWII economic order.