“Intelligence augmentation decreases the need for specialization and increases participatory complexity.”

Jamais Cascio, author and futurist

 

This JAMA Research Letter focuses on a very relevant topic of how clinicians, in this case dermatologists, are accommodating artificial as well as augmented intelligence (AuI) in their practices.

A cross-sectional survey was performed by the American Academy of Dermatology (AAD) Task Force on Augmented Intelligence of over 3,000 fellows in July 2020. Only 121 dermatologists completed the survey (3.9%). It is somewhat disconcerting that such a low percentage of those surveyed responded to the survey on artificial and augmented intelligence.

The results showed that only 46% thought artificial intelligence would positively influence their practice, 64% felt that way for augmented intelligence. Interestingly, 76% of respondents stated that they would be more likely to perform a biopsy that was thought not to be suggestive of cancer IF an AI tool indicated the lesion to be possibly malignant while only 8% would be less likely to perform a biopsy of a clinically suspicious lesion IF the AI tool indicated that the lesion is benign. In other words, the dermatologists are 10 times more likely to biopsy a lesion that AI thought may be malignant compared to foregoing a biopsy of a benign one if their clinical suspicion differs from the AI.

The top potential benefits of AI tools for skin cancer screening are more efficient triage and improved health care access, while the top risks include loss of patient follow-up owing to using AI without clinician support, clinician loss of control to AI, and human deskilling.

In addition, the group felt that the top strengths of AI included patient motivation to seek skin cancer diagnosis and therapy, more objective diagnosis, and more convenient diagnosis while the top weaknesses included inability to perform total body skin examinations, lack of creativity owing to algorithmic limitations, and a lack of social contact.

The group also felt that the greatest potential challenge of implementing AI or AuI in their practice is disruption of human physician-patient relationship. Overall, 74% of the dermatologists surveyed are willing to use an AI tool with similar accuracy to a human dermatologist to help monitor skin conditions.

This research letter is especially timely given the Google announcement of the Dermatology Assist tool to help educate the public on skin conditions. The survey indicated that the dermatologists are receptive to assistance from AI tools but value the human-to-physician relationship and accuracy.

It is possible also that the low response rate is indicative of bias towards users or advocates for AI so that the prevailing attitude of AI may be considerably less positive.

The full article can be read here