Dr. Foteini Agrafioti is the senior vice-president for data and artificial intelligence and chief science officer at Royal Bank of Canada.
Artificial intelligence continues to be one of the most transformative technologies affecting the world today. It’s hard to imagine an aspect of our lives that is not impacted by AI in some way. From assisting our web searches to curating playlists and providing recommendations or warning us when a payment is due, it seems not a day goes by without assistance from an AI insight or prediction.
It is also true that AI technology comes with risks, a fact that has been well documented and is widely understood by AI developers. The race to use generative AI in business may pose a risk to safe adoption. As Canadians incorporate these tools into their personal lives, it has never been more important for organizations to examine their AI practices and ensure they are ethical, fair and beneficial to society. In other words, to practice AI responsibly.
So, in an era where AI is dominating, where do we draw the line?
When everyone is pressing to adopt AI, my recommendation is do so and swiftly, but start with deciding what you are NOT going to use it for. What is your no-go zone for your employees and clients? Devising principles should be a critical component of any organization’s AI ecosystem, with an organized and measured approach to AI that balances innovation with risk mitigation, aligns with human values and has a focus on well-being for all. This is particularly pronounced in highly regulated sectors where client trust is earned not just through results but also with transparency and accountability.
Encouraging this conversation early may help inform important decisions on your approach. At RBC we had the opportunity to discuss these concepts early in our journey, when we hired AI scientists and academics in the bank and were faced with the opportunity to push boundaries in the way we interact with our clients. A key decision that we made a decade ago was that we wouldn’t prioritize speed over safety. This fundamental principle has shaped our approach and success to date. We have high standards for bias testing and model validation and an expectation that we fully understand every AI that is used in our business.
This cannot just be a philosophical conversation or debate.
The plan to build AI according to a company’s ethos needs to be pragmatic, actionable and available for education and training to all employees. Organizations of all sizes and in all sectors should develop a framework of responsible AI principles to ensure the technology minimizes harm and will be used to benefit clients and the broader society. For large companies, this can manifest in how they build their models, but even small businesses can establish ground rules for how employees are permitted to use for example, off-the-shelf GenAI tools.
RBC aims to be the leader in responsible AI for financial services and we continue to innovate while also respecting our clients, employees, partners, inclusion and human integrity. The two can go hand-in-hand when using AI responsibly is a corporate priority. Our use of AI is underscored by four pillars: accountability, fairness, privacy and security and transparency.
From senior leaders to the data analysts working on the projects, we put the clients at the centre of our AI work by asking the right and strategic questions to ensure it aligns with RBC’s values. When developing AI products or integrating with a third-party vendor, organizations should consider adopting policies and guidelines to understand third-party risks such as scheduled audits, in-depth due diligence, product testing, user feedback, validation and monitoring to ensure AI systems are developed and deployed in a responsible and transparent manner.
Employee training is an equally important but often overlooked element of responsible AI, ensuring employees know not only how to prompt effectively, but also how to supervise the outputs. This includes guidance on the use of external tools for work-related products, the kinds of information that can be uploaded or included in prompts and use cases that have the most to gain from what AI can deliver safely.
At RBC, all leaders and executives who have access to GenAI tools complete training that addresses responsible AI considerations, governance and decision-making to enable informed oversight and guidance on responsible AI initiatives. Employees who are granted access to GenAI support tools need to take general training, which includes an introduction to responsible AI principles, concepts and ethical frameworks. AI models are subjected to thorough testing, validation and monitoring before they are approved for use. This is not a risk-averse culture. It’s a culture that honours the existing partnership it has with its clients.
This is not once and done. Ongoing governance to review and assess the principles remains important. RBC’s responsible AI working group continues to ensure the principles remain relevant, address business and stakeholder needs and drive linkages between the organization’s responsible AI principles, practices and processes.
Another approach is to partner with leaders including educational institutions and think tanks in the AI space to provide strategic opportunities to help drive responsible AI adoption. RBC recently partnered with MIT to join a newly formed Fintech Alliance, focused on addressing pressing matters around the ethical use of AI in financial services.
This type of fulsome approach helps recruit in the AI space from leading educational programs globally, which brings together divergent opinions and different work experience, creating a diverse workforce which creates an environment where collaboration, learning and solutions thrive.
As modern AI systems advance, navigating platform ethics can be complex. Implementing a strong responsible AI framework, inclusive of principles and practices, will ensure challenges are addressed ethically and AI systems are adopted responsibly.
This column is part of Globe Careers’ Leadership Lab series, where executives and experts share their views and advice about the world of work. Find all Leadership Lab stories at tgam.ca/leadershiplab and guidelines for how to contribute to the column here.