New Research Identifies “Agentic Loafing,” the Hidden Risk of Clinicians Handing Decision Authority to AI

HERNDON, VA, (July 27, 2026) — A forthcoming study in the journal Risk Analysis introduces “agentic loafing,” a term for what happens when clinicians stop using AI as a tool and start simply ratifying its judgments. This research identifies three institutional drivers that normalize AI delegation in healthcare and offers a preliminary diagnostic tool that turns unmanaged bets into something measurable.

Key Findings:

• Three institutional drivers normalize AI delegation. Performance culture conformity creates the pressure to delegate. Structural isolation of hierarchical responsibility creates the opportunity to offload responsibility without consequence. Legitimization through quantification provides the rationalization that makes AI delegation feel justified.
• However, accountability does not transfer. Classical delegation theory assumes monitoring, punishment and shared objectives keep an agent honest. AI delegation thrives in the absence of all three. In healthcare settings, clinicians can blame the algorithm for bad outcomes and claim credit for good ones. The researchers call the result “risk evaporation,” the systematic disappearance of accountability when an AI-assisted decision fails, and no single party is responsible for the outcome.
• The risk is now measurable. The study delivers a preliminary diagnostic tool that scores a clinical unit’s risk of agentic loafing from 0 to 5 . The radar profile shows which dimension drives the score, so units can target interventions rather than apply blanket fixes.
• The warning inverts the usual assumption about AI risk. The study argues that the most dangerous AI systems are not those that fail unpredictably, but instead, they are the ones that function just reliably enough to quietly erode human judgment without ever triggering an alarm.

Methodology:

The team analyzed 11 expert podcast episodes featuring healthcare administrators, clinicians, regulators and AI researchers. Findings were validated through secondary analysis of eight white papers from authoritative sources including the OECD, the World Economic Forum and McKinsey & Co. Cross-source convergence between podcast-derived themes and white paper evidence ranged from 80% to 90%.

Authors Available for Interview:

Helmi Issa, Ph.D., corresponding author, associate professor in operations management and decision science, ESSCA School of Management, France

Fabio James Petani, Ph.D., associate professor in digital management, Burgundy School of Business, Dijon, France

Dejan Glavas, Ph.D., associate professor in AI and sustainability, ESSCA School of Management, France

“Agentic Loafing: An AI Decision Delegation Risk,” Risk Analysis.

###

About Risk Analysis

Risk Analysis is a peer-reviewed journal, publishing original research on the assessment and management of risks across disciplines including public health, engineering, environmental science, social science and policy. Risk Analysis, founded in 1980, is published by Wiley on behalf of the Society for Risk Analysis.  

Media Contact 

Anya Grondalski | Anya@bigvoicecomm.com