Patient Perceptions on the Use of Artificial Intelligence in Creating Clinical Research Documents: Survey Study.

Background: The use of generative artificial intelligence (AI) by pharmaceutical companies and other organizations for preparing patient-facing documents reporting results of clinical research is becoming more common. This raises concerns about whether the accuracy and quality of these documents could be affected, as well as the potential impact on patient perceptions and trust. Accurate and trustworthy information is critical to health care decision-making. Little is known about patient perceptions of AI-generated content. Objective: This study aimed to better understand patient experience and familiarity with AI, their resulting confidence in the abilities of AI, and their trust in the use of AI by research organizations to generate clinical research documents. Methods: An online survey was conducted using an online health care panel of patients in Europe and the United States to assess familiarity with AI, trust in organizations reporting on research, and trust in the use of AI to prepare clinical research documents. The survey also asked directly about the importance of human involvement and of transparency in disclosing AI use. Results: A total of 1010 respondents completed the online survey. About half of respondents were from the United States and half from Europe. Survey results showed that 63.6% (642/1010) of respondents had used AI before with 74.9% (756/1010) reporting being "Somewhat" or "Very" familiar with AI. AI use was influenced by country, gender, education level, race/ethnicity, and clinical trial experience. Higher familiarity with AI was observed among younger participants. Respondents were generally confident in the capabilities of AI, as more than half believed AI use would reduce grammar and data errors. Trust in clinical trial documents generally increased with greater human oversight, as trust was lowest for documents created by AI with no human involvement (12.3% "A lot" of trust, 124/1010) and highest for documents created by humans without AI (39.1% "A lot" of trust, 395/1010). 95.0% (959/1010) of respondents considered human involvement in clinical trial document review as "Very important" or "Somewhat important." The majority (62.4%, 630/1010) of respondents felt it was "Very important" for pharmaceutical companies and academic institutions to be transparent about their use of AI in public-facing documents. Transparency was considered more important among respondents in the United States and the United Kingdom compared to those in the European Union. Conclusions: The survey results reveal high familiarity with, and confidence in the capabilities of, AI. Despite this confidence, respondents emphasized the need for human involvement in the creation of clinical trial documents and the importance of disclosing AI usage, underscoring the critical role of human oversight in maintaining patient trust. Transparent integration of AI with deliberate human involvement remains essential to ensure trust in patient-facing documents.