Reflections on Best Practices: Building a Stronger Research Ethics Board and Addressing Implementation Challenges in a Rapidly Growing University in Canada

DESCRIPTION OF POSTER

"The emergence of Artificial Intelligence (AI) research in healthcare poses a plethora of ethical challenges for REBs. Due to the rapid growth and adoption of AI research, recognition of and management of ethical issues are lagging due to a lack of adaption of research policies and normative guidelines. In this poster presentation, we will provide an overview of and comparison between conventional ways we utilize and apply the ethical principles from the TCPS2 and how they ought to be applied to AI-focused research. This poster presentation will explore how to incorporate the TCPS2 principles of respect for persons, concern for welfare and justice into AI-focused research. Some examples include considering how respect for persons can be applied to large volumes of data, as well as how to manage AI solutions that involve black-box decision-making, where explainability and truly informed consent may give REB reviewers pause. The TCPS2 principle of concern for welfare requires an assessment of the potential and actual risks and benefits imposed on participants, populations and communities. Early engagement with those most impacted by AI solutions ought to occur to help determine the risks and benefits, especially regarding trust and data security and ownership in the digital age. Lastly, the TCPS2 principle of justice reflects fairness and equity in practice. AI solutions have enormous potential to impact individuals, as well as communities in discriminatory and unjust ways on an unprecedented scale. Although concerns related to equity existed prior to AI-solution research, they now have the potential for exponential harms, especially to certain groups. Questions related to accountability and responsibility can be harder to address when looking at AI research. Clear lines of oversight, and closer attention much be paid when exploring these concerns and addressing them along the AI research lifecycle. Are there new principles that ought to be considered in this emerging lens or will the TCPS2 principles suffice, albeit with new ways of applying them in the novel AI context? Our presentation will seek to provide guidance on these questions."

Reflections on Best Practices: Building a Stronger Research Ethics Board and Addressing Implementation Challenges in a Rapidly Growing University in Canada

DESCRIPTION OF POSTER

"The emergence of Artificial Intelligence (AI) research in healthcare poses a plethora of ethical challenges for REBs. Due to the rapid growth and adoption of AI research, recognition of and management of ethical issues are lagging due to a lack of adaption of research policies and normative guidelines. In this poster presentation, we will provide an overview of and comparison between conventional ways we utilize and apply the ethical principles from the TCPS2 and how they ought to be applied to AI-focused research. This poster presentation will explore how to incorporate the TCPS2 principles of respect for persons, concern for welfare and justice into AI-focused research. Some examples include considering how respect for persons can be applied to large volumes of data, as well as how to manage AI solutions that involve black-box decision-making, where explainability and truly informed consent may give REB reviewers pause. The TCPS2 principle of concern for welfare requires an assessment of the potential and actual risks and benefits imposed on participants, populations and communities. Early engagement with those most impacted by AI solutions ought to occur to help determine the risks and benefits, especially regarding trust and data security and ownership in the digital age. Lastly, the TCPS2 principle of justice reflects fairness and equity in practice. AI solutions have enormous potential to impact individuals, as well as communities in discriminatory and unjust ways on an unprecedented scale. Although concerns related to equity existed prior to AI-solution research, they now have the potential for exponential harms, especially to certain groups. Questions related to accountability and responsibility can be harder to address when looking at AI research. Clear lines of oversight, and closer attention much be paid when exploring these concerns and addressing them along the AI research lifecycle. Are there new principles that ought to be considered in this emerging lens or will the TCPS2 principles suffice, albeit with new ways of applying them in the novel AI context? Our presentation will seek to provide guidance on these questions."