AI's competence erodes institutional judgment and tacit knowledge (socialeurope.eu)

• AI's efficiency gains simultaneously reduce human capacity for the task
• Institutions rely on tacit knowledge built over decades, which AI erodes
• Current regulations like the AI Act miss cumulative erosion by assessing systems individually
• Proposed solutions include a dependency index and defended boundaries

"The article argues that the primary risk of artificial intelligence is not its failures but its competent performance, which simultaneously provides efficiency and erodes human judgment. Each time AI performs a task, the human capacity for that task is left unexercised. This is especially dangerous for institutions like ministries and regulators, which rely on accumulated tacit knowledge built over decades. AI replaces junior roles that are essential for developing future judgment, and current regulations like the EU AI Act assess systems individually, missing the cumulative erosion. The author proposes measuring a dependency index and establishing defended boundaries for AI use."