[IN-PERSON/LIVESTREAMED] Plenary Session: Ethical Considerations for Using Generative Artificial Intelligence (AI) in Human Subjects Research‐‐Implications for IRB Regulation

This plenary session will be held in-person, livestreamed for the Virtual Meeting, and recorded for on-demand access by attendees registered for PRIMR23 on this day.

Generative AI is a rapidly advancing technology that allows computers to learn from training data and generate “content” that are similar to those produced by humans. Chat GPT is one example that has gained increasing public recognition and use. Generative AI has the potential to revolutionize many aspects of human subjects research, from generating interview questions and protocol or consent materials, to creating synthetic data. However, the use of Generative AI in human subjects research raises a host of ethical questions that must be addressed—many we cannot even fathom because we know so little about the technology. For example, how can we ensure that a consent generated by an AI is accurate and sufficiently understandable to research participants? Should we be using AI generated content for consents? What are the potential risks and benefits of using AI-powered chatbots in health interventions? Can the Generative AI be trusted to interact independently with research subjects? How can the researcher be sure the Generative AI won’t go “off the rails” like we have seen every current Generative AI do, for example, telling someone seeking mental health services to harm themselves? What safeguards can researchers put in place when they use Generative AI to interact with research participants? How can we prevent Generative AI from perpetuating bias in data analysis or in generating biased materials and recommendations for participants? Moreover, the use of Generative AI in research presents a unique challenge for IRBs tasked with regulating this research. IRBs will need to understand the capabilities and limitations of Generative AI in order to make informed decisions about the ethical implications of using this technology in research. But, where do they start to learn?