Table 3.

Summary of patient data addressing the themes of confidence in the capabilities of AI, its accountability and transparency, and demographic features associated with these views

Authors, yearConfidenceAccountabilityPatient demographics influencing positive attitudesPatient demographics influencing negative attitudes
Müller et al, 2021 11All participants saw potential for AI to improve care and evidence of maintaining standards
Recognised potential for AI error
Concern that malfunctions of AI might not be detectable or correctable
Concern regarding introduction of potential AI bias and lack of accountability by dentists
Data protection issues
Expectation of validation and regulatory approval
Expectation of understanding reasoning behind AI
Digital literacy
Previous exposure to AI
Previous experience of error
Seeing a potential for improving care
Lennartz et al, 2021 1468% believed AI decisions should be verified by physician
79.5% disagreed with AI operating without decisions being checked by physician
34% believed AI would mostly make the correct diagnosis, 28% undecided, 15% disagreed
67% believed physician's decision should be favoured over AI if they disagreed (25% undecided)Familiarity with AI technologyAcceptability of AI declined with respect to greater disease severity
Ongena et al, 202112Moderate scepticism regarding AI
78% agreed/strongly agreed that physician is necessary to verify AI interpretation of mammograms
42% opposed AI selecting cases requiring physician interpretation
Ambiguous views regarding whether physicians or developers are responsible for errorsAcademic attainmentCharacteristics favouring personal interaction
Ongena et al, 202015Moderately sceptical of diagnostic accuracy of AIPatients indicated need to be informed on all aspects of diagnostic process
Desire for ethical and legal framework within which AI operates
Academic attainmentCharacteristics favouring personal interaction and dis-inclination to change
York et al, 202013Moderate confidence in AI compared with high confidence in ability of physicians95.4% of participants selected physician's opinion in event of disagreement with AIStudent
Academic attainment
Weak association with technology usage
Inability to work
Jonmarker et al, 20199Faster results identified as potential benefits of AI
High levels of trust in AI expressed
67% would be satisfied with immediately available AI-only report
59% would have confidence in report
97% would have confidence in combined AI/physician report>12 years of formal education
Familiarity with technology
History of chronic disease
Age not identified as factor
Haan et al, 201910Limited patient awareness of AI and of radiology
Patients sceptical about performance of AI and prefer AI as a second opinion
Restricted evaluation by AI seen as potential source of incorrect diagnosis
Expectation of scientific validation
Evidence of efficacy is responsibility of ‘scientific community’
Not evaluatedNot evaluated