R0044/2026-04-01/Q002/S02
WebSearch — Healthcare AI errors, false confirmation, and clinical decision support harm
Summary
| Field |
Value |
| Source/Database |
WebSearch |
| Query terms |
"AI agreeable output harm incident reports near-miss healthcare clinical decision support wrong diagnosis" + '"alert fatigue" "commission error" AI clinical decision support system design requirements preventing confirmation bias' |
| Filters |
None |
| Results returned |
20 |
| Results selected |
1 |
| Results rejected |
19 |
Selected Results
Rejected Results
| Result |
Title |
URL |
Rationale |
| S02-R02 |
Various healthcare AI error articles (19 results) |
Various |
Included: general AI diagnostic error articles, malpractice liability discussions, alert fatigue studies, and bias recognition reviews. Most address AI being wrong rather than AI agreeing with incorrect user assumptions. The false confirmation study from Nature Communications was the only source specifically addressing the agreement mechanism. |
Notes
Healthcare AI error literature is extensive but primarily focused on AI producing incorrect output (misdiagnosis, biased recommendations) rather than AI confirming incorrect human assumptions. The false confirmation study from Nature Communications was the key discriminating find — it specifically addresses the mechanism where AI agreement with an incorrect clinician hypothesis leads to harm.