48 - Building a National Working Group in Response to New AI/ML Ethical Challenges

Background:  Exponential growth in AI/ML/Big Data human subjects research has surpassed the existing research guidance and best practices. Since this is a new frontier, IRBs are struggling to build guidance in this space. We spearheaded a collaborative working group to address these challenges to develop shared deliverables.

Program

Description:  We formed a collaborative working group with 14 other institutions across the nation. The larger group identified priority areas of focus, then we divided the larger group into smaller working groups - ethics, data quality, identifiability, IRB/HRPP scope, and group charter development.

Our larger group meets monthly to discuss the newest developments in the field and to provide status updates from the smaller subgroups. The subgroups are tasked with developing guidelines and tools for large-group review, which will culminate in thoughtful guidance made accessible to all institutions and entities.

Program

Assessment:  In bringing together our collective minds across the nation, we realized that this issue transcends individual institutions and disciplines and requires an interdisciplinary collaborative approach to identify best practices.
This approach has worked to the extent that we have come together, identified the issues, broken into subgroups, and are working towards our deliverables.
Some of the challenges include a quickly changing landscape as it relates to research. This requires expertise that is outside of most HRPPs and thus we have sought training from both experienced researchers and leaders in academia.

Limitations: One limitation is that each institution is at a different stage in developing practices to address AI/ML concerns. A second limitation is the group size of 15 institutions, offering only 15 perspectives, but decades of IRB/HRPP experience and extensive professional networks within. The most significant limitation has been the lack of knowledge and information available on AI/ML/Big Data risks, capabilities, unknowns, and how that will translate into human subjects’ research.

Discussion:  Other institutions could adopt this method to address unique considerations of emerging technologies and other topics that benefit from collaboration and the sharing of ideas across institutions. This specific area of research appears to be an ever-evolving field that will necessitate revisiting guidelines, best practices, etc. The collaborative working group is an already established group that will likely continue into the future, making adjustments along the way.