Ctrl + Alt + Conceive: fertility awareness in the age of Artificial Intelligence, how do large language models compare?

Abstract

Technology continues to change how we manage our health, and recent breakthroughs in Artificial Intelligence have increased the adoption of Large Language Models (LLMs) in healthcare. Since the launch of ChatGPT, LLMs have been increasingly used for health information; this study, therefore, aimed to qualitatively assess fertility information provided by LLMs. Content generated by four LLM platforms: ChatGPT, Gemini, Copilot, Perplexity, were analysed comparatively. Thirty-seven prompts were generated, covering five topics: menstrual cycle, conception, risk factors, assisted reproductive technologies and age-related fertility decline. Prompts were analysed for concordance, comprehensibility and conciseness. Safety warnings for all platforms were recorded. LLM platforms generally provided concordant answers for menstrual cycle, conception, and risk factors. However, content on assisted reproductive technologies was the least accurate. Perplexity provided the highest number of strongly-concordant and poorly-concordant responses. Comprehensibility was similar across platforms. ChatGPT was the most concise. Not all platforms provided warning or safety messages regarding potential inaccuracies. LLMs present an opportunity to expand access to fertility and reproductive health information not only for individuals and patients, but also for clinicians, researchers, educators, charities, reproductive health organisations and policymakers. Nevertheless, attention must be paid to the quality of information generated in order to ensure that professionals have accurate guidance, and that individuals can access quality information to help achieve their desired fertility and reproductive health intentions.

Ähnliche Arbeiten