AI Decommissioning Leaves Staff With Mixed Feelings — and No Good Playbook

Pulling a chest X-ray AI triage tool caused both operational relief and perceived clinical loss among NHS radiology staff. Post-decommissioning, 70% still saw patient benefit (vs 71% pre-implementation), but logistical burden rose vs late-use. Authors propose a 3-phase off-board…

• Anonymous survey at a multi-site NHS trust, 2 months post-decommissioning; 40/187 responses (21.4%), comparable to earlier rounds but reporting-staff response was very low (5/40), limiting role-specific conclusions.
• Perceived logistical burden was higher after withdrawal (35%) than during late implementation (26.4%) but lower than early implementation (51.7%); qualitative themes included loss of pathway efficiency, operational relief, and emotional labour during AI-enabled escalations.
• Key limitation: single-trust retrospective survey with small n and low reporting-staff response — findings may not generalise beyond similar NHS settings or this tool type.

RadPigeon summaries are original and for information only. They are not clinical advice.