Agentic Loafing: An AI Decision Delegation Risk

{"Artificial":[0],"Intelligence":[1],"(AI)":[2],"is":[3],"increasingly":[4],"making":[5],"decisions":[6,102],"in":[7,83],"healthcare,":[8],"not":[9,155],"just":[10,164],"supporting":[11],"them.":[12],"This":[13,86],"research":[14,91,116],"introduces":[15],"\\"agentic":[16,129],"loafing\\"":[17,130],"to":[18,30,167],"describe":[19],"the":[20,95,149],"intentional":[21],"yet":[22],"unexamined":[23],"delegation":[24,69,73,81],"of":[25,35,41,59,98],"decision-making":[26],"authority":[27],"from":[28,131],"clinicians":[29],"AI":[31,52,80,152],"systems.":[32],"Using":[33],"netnography":[34],"professional":[36],"podcasts":[37],"and":[38,61,77],"secondary":[39],"analysis":[40],"gray":[42],"literature,":[43],"this":[44,90,113,115],"study":[45],"uncovers":[46],"three":[47],"institutional":[48],"drivers":[49,66],"that":[50,127,157,162],"normalize":[51],"delegation:":[53],"performance":[54],"culture":[55],"conformity,":[56],"structural":[57],"isolation":[58],"responsibility,":[60],"legitimization":[62],"through":[63],"quantification.":[64],"These":[65],"invert":[67],"classical":[68],"theory.":[70],"Where":[71],"traditional":[72],"assumes":[74],"monitoring,":[75],"punishment,":[76],"shared":[78],"objectives,":[79],"thrives":[82],"their":[84],"absence.":[85],"inversion":[87],"produces":[88],"what":[89],"terms":[92],"\\"risk":[93],"evaporation\\":":[94],"systematic":[96],"disappearance":[97],"accountability":[99],"when":[100],"AI-assisted":[101],"fail,":[103],"leaving":[104],"no":[105],"single":[106],"party":[107],"responsible.":[108],"To":[109],"help":[110],"organizations":[111],"manage":[112],"risk,":[114],"also":[117],"develops":[118],"a":[119,136,142],"\\"preliminary\\"":[120],"diagnostic":[121],"tool,":[122],"with":[123],"its":[124],"intersectional":[125],"interventions,":[126],"transforms":[128],"an":[132,175],"unmanaged":[133],"bet":[134],"into":[135],"measurable":[137],"risk":[138,147],"score.":[139],"From":[140],"this,":[141],"critical":[143],"implication":[144],"follows":[145],"for":[146],"analysis:":[148],"most":[150],"dangerous":[151],"systems":[153],"are":[154],"those":[156,161],"fail":[158],"unpredictably":[159],"but":[160],"function":[163],"reliably":[165],"enough":[166],"quietly":[168],"erode":[169],"human":[170],"judgment":[171],"without":[172],"ever":[173],"triggering":[174],"alarm.":[176]}

Authors

Institutions

Publication Details

Journal
Risk Analysis
Published
2026-07-27
DOI
https://doi.org/10.1111/risa.70306
Primary Topic
Artificial Intelligence in Healthcare and Education
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
article

Agentic Loafing: An AI Decision Delegation Risk

Fabio James Petani, Dejan Glavas, Helmi Issa
Risk Analysis
Artificial Intelligence in Healthcare and Education
article

Agentic Loafing: An AI Decision Delegation Risk

Fabio James Petani, Dejan Glavas, Helmi Issa
article en

Abstract

Artificial Intelligence (AI) is increasingly making decisions in healthcare, not just supporting them. This research introduces "agentic loafing" to describe the intentional yet unexamined delegation of decision-making authority from clinicians to AI systems. Using netnography of professional podcasts and secondary analysis of gray literature, this study uncovers three institutional drivers that normalize AI delegation: performance culture conformity, structural isolation of responsibility, and legitimization through quantification. These drivers invert classical delegation theory. Where traditional delegation assumes monitoring, punishment, and shared objectives, AI delegation thrives in their absence. This inversion produces what this research terms "risk evaporation": the systematic disappearance of accountability when AI-assisted decisions fail, leaving no single party responsible. To help organizations manage this risk, this research also develops a "preliminary" diagnostic tool, with its intersectional interventions, that transforms "agentic loafing" from an unmanaged bet into a measurable risk score. From this, a critical implication follows for risk analysis: the most dangerous AI systems are not those that fail unpredictably but those that function just reliably enough to quietly erode human judgment without ever triggering an alarm.

Risk AnalysisVol. 46(8)
Université de Bourgogne (FR), Université Bourgogne Franche-Comté (FR), ESSCA School of Management (FR)
Peace, Justice and strong institutions
Openalex Percentile: Top 12%
Artificial Intelligence in Healthcare and Education
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.