Pediatric physicians sound the alarm: GenAI is a cause for concern

Glenna Crooks, PhD and Paul Hambly, EVP of Supply at Toluna

AI is transforming industries across the board, and healthcare is no exception. This is the fifth in a series of articles reviewing physicians’ perspectives on GenAI, addressing the views of pediatricians, who are consistently among the most cautious voices on the risks of GenAI.

Pediatric cautions

Across nearly every risk category measured in our survey, physicians caring exclusively for children expressed significantly higher concerns than their colleagues. The data suggests something profound about how physicians view their responsibility when caring for society’s youngest and most vulnerable members. The pattern is unmistakable: across every dimension, pediatric physicians’ GenAI risk assessment is the most stringent – for multiple reasons:

Why children change everything

Our survey didn’t directly ask about protective instincts, but the data pattern suggests pediatric physicians may indeed feel a heightened responsibility for their young patients. This isn’t surprising when you consider the unique vulnerabilities children face in healthcare settings:

• Consent and agency: Children cannot consent to their care or understand the implications of GenAI assistance. Parents and physicians bear the full weight of decision-making.
• Developmental considerations: Children’s physiology, psychology, and treatment responses differ dramatically from those of adults. GenAI systems trained primarily on adult data may not safely translate to pediatric care.
• Long-term consequences: Errors in pediatric care can compound over time. A misdiagnosis at age 5 has different implications than one at age 65.
• Family dynamics: Pediatric physicians must navigate patient care and family communication, adding complexity to the transparency requirements surrounding the use of GenAI.
• Privacy violations: Disclosure of children’s medical data represents decades of potential risks that could impact patients well into adulthood.
• Training data bias:  Biased GenAI recommendations during childhood could set trajectories that perpetuate inequities for decades.

Developer accountability for harm

Pediatricians are more likely to hold developers accountable for any harm caused by GenAI (30%) compared to adult-only physicians (23%) and mixed adult-pediatric physicians (23%).

Coming next:

Our next article will delve into the transparency dilemma physicians face: acknowledging that patients should be informed about the use of GenAI in their care while fearing the consequences of disclosing this information.