A Snapshot from Our Working Groups: Q&A with Allison Cohen on the Importance of Independent Review
Q1: What do you determine is the main function of independent review of AI Development? Why is it important?
Allison Cohen (AC): Independent review of AI is important for three reasons. First, it entails meaningful AI obligations. Second, it ensures those obligations are standardized across companies and third, it generates accountability for those obligations. Each of these developments are breakthroughs in the realm of ethical AI, which has previously been characterized by weak commitments, uneven applications and a lack of responsibility.
Q2: What is the role of community in independent review?
(AC): In the Independent Review Guidelines, the community advocate plays the role of representing the needs and concerns of their stakeholder group to the ethical review committee. It’s critically important that the community advocate is meaningfully included in this process because, otherwise, the technology’s ethical implications cannot be sufficiently considered. The community advocate thereby plays the role of creating a more robust ethical review process, which may otherwise fail to account for the true ethical implications of the technology.
When choosing a community advocate, it is important that the individual be a leader within their community and can sufficiently represent the needs and concerns of their constituents.
Q3: Who are responsible AI guidelines for?
(AC): Responsible AI Guidelines are for any organization undergoing AI research and development, whether the technology is being sold directly to the consumer or being used by another corporate or public sector entity. Wherever the AI is being used, there is ample risk associated with the entire lifecycle of AI design and deployment. As such, all organizations developing this technology must ensure they conduct a responsible AI review.
Q4: What can AI practitioners do to ensure that they are facilitating independent review for their systems? What can regulators do?
(AC): AI practitioners can make sure they properly capture their processes when developing AI technology. This notation will be used to support a more robust ethical AI review process. Furthermore, AI practitioners can promote a culture of ethical AI development within their organization by raising the importance of ethical AI development with their managers and asking for training.
It is critical that regulators play an active role in ensuring ethical review processes are required in addition to being robust, meaningful and effective. Regulators are often concerned with the optics of regulating ethical AI given the widespread misunderstanding that regulation inhibits economic development. However, regulators must actively dispel this myth as it creates significant barriers for ethical AI regulation and it actively undermines the quality of the AI being developed.
Allison Cohen is the Applied AI Projects Lead at Mila- Quebec Artificial Intelligence Institute. In this role, Allison manages the development and deployment of socially beneficial AI tools. Allison is passionate about ethical AI development and has consulted Global Affairs Canada on ways of promoting human rights respecting AI internationally.