Do people prefer AI-generated patient educational materials over traditional ones?

Abstract

Objective

This study aimed to assess people’s preference between traditional and Artificial Intelligence (AI)-generated colon cancer staging Patient Education Materials (PEMs).

Methods

We assessed preference among patients and companions being seen for a non-cancer diagnosis at the UT Health Austin Colon and Rectal Surgery Clinic. Participants were blinded to the study concept of AI and generation method of PEMs (Traditional: National Cancer Institute and the American Cancer Society; AI-generated: ChatGPT and Google Bard). Participants completed an anonymous demographic survey and ranked their PEM preference for randomized, de-identified materials. We hypothesized that there would be no difference in preference between the AI-generated and traditional PEMs. This study is IRB approved (STUDY00005203).

Results

35 participants were recruited, ages 27–79 years (mean 46; SD 14 years) and 57 % female. 54 % were four-year college graduates. Among the participants, 51 % ranked the traditional PEMs as most preferred, while 49 % ranked the AI-generated PEMs as most preferred (p = NS).

Conclusion

People do not have a dominant preference for AI-generated or traditional PEMs.

Practice implications

Further investigation on the use of AI for PEM generation is warranted. AI may be capable of generating PEMs that incorporate the strengths of traditional PEMs while also reducing the required resources for PEM generation.