Work

Response to: Critical Evaluation of Generative Text-to-Image Artificial Intelligence Models in Plastic Surgery: Balancing Promise and Precision

Year: 2025

Type: article

Abstract: We sincerely appreciate the thoughtful letter by Drs Ozmen and Schwarz in response to our recent publication, “Utilizing Generative Text-to-Image AI Models to Explore Race, Gender, and Age in Plastic and Aesthetic Surgery.”1,2 Their commentary highlights critical issues surrounding the application of generative artificial intelligence (AI) in clinical and aesthetic contexts and underscores the broader implications of such technologies in shaping patient expectations and practice workflows.1 We would like to address and expand on several key points raised in their letter. We fully agree that DALL-E3 (OpenAI, San Francisco, CA), Midjourney (Midjourney Inc., San Francisco, CA), and Adobe Firefly (Adobe Inc., San Jose, CA) are not purpose-built for medical use. In fact, our manuscript explicitly acknowledged this limitation and was intended not as an endorsement of these tools for clinical decision making, but as an exploration of how their outputs, which are already accessible to the public, could influence patient perceptions.2-4 In our study, we aimed to critically assess current biases and representational patterns in these systems, particularly given their potential to shape aesthetic norms outside clinical settings.2 Additionally, Drs Ozmen and Schwarz correctly point out the lack of anatomical precision in AI-generated images, and we share their concern.2-4 Our study emphasized that these tools are not substitutes for clinical visualization or preoperative simulation.2 We stated explicitly that these models are unvalidated for surgical planning, and our goal was to assess their aesthetic representations through the lens of diversity and inclusivity, not to evaluate their anatomical accuracy.2 Moreover, their reinforcement of concerns about westernized beauty ideals and gender stereotypes further validates our findings. We appreciate their emphasis on how these biases can be internalized by patients, affecting expectations and satisfaction. This aligns with our discussion on how such models can inadvertently propagate limited or unrealistic standards and how critical engagement by clinicians is needed to contextualize these outputs for patients. In conclusion, we wholeheartedly support the authors' call for active involvement by the plastic surgery community in the development of specialized AI tools. As we noted in our discussion and future directions, a collaborative approach, including surgeons, ethicists, engineers, and patient advocates, is essential to building inclusive, medically relevant, and clinically useful models. We are encouraged by their emphasis on the need for anonymized datasets, standardized evaluation tools, and ethical frameworks in shaping the next generation of AI technologies in plastic surgery. We thank Drs Ozmen and Schwarz for their insightful comments and for underscoring the urgency of responsible AI integration in aesthetic medicine. Their letter enriches the ongoing discourse, and we hope that our collective efforts continue to guide the development of more equitable, transparent, and clinically aligned AI tools. The authors declared no potential conflicts of interest with respect to the research, authorship, and publication of this article. The authors received no financial support for the research, authorship, and publication of this article.

Source: Aesthetic Surgery Journal

DOI: 10.1093/asj/sjaf106