Artificial intelligence (AI) continues to become increasingly widespread and organizations are ever becoming more reliant on AI systems for critical decision-making. With this, there is a need for a framework that will guide them in ensuring that they are able to implement their AI initiatives responsibly.
Recognizing this need, Union Bank of the Philippines(UnionBank) released a report that will act as a guide on how responsible AI can be enforced on the greater industrial level. The contents of the report were based on a collaborative initiative with the Gradient Institute called Project AI Trust, which has enabled the Bank to consider and implement responsible AI practices for its automated systems.
Through the report, UnionBank proposed four measures that the industry can adopt to ensure responsible AI governance: broadening the scope of existing AI governance; providing staff with additional support and incentives; defining new functions with specific governance roles; and creating documentation & resources to support governance activities.
Another key recommendation is the creation of an Algorithmic Ethics Committee that will ensure the safety of the AI processes, and that system owners are observing ethical business practices. This committee should compose of members from different business units in the organizations, as well as external specialists or domain experts as needed.
“While AI can provide numerous benefits that can uplift lives, it is also not perfect, and safeguards must be put in place to ensure that AI systems are used for the exact purposes set by their organizations,” said Chief Data & Artificial Intelligence OfficerDr. David Hardoon.
In addition, the report also proposed a 4-stage system life cycle where the Algorithmic Ethics Committee should conduct reviews on the development of the AI system. In chronological order, these stages are: design review, which will focus on the system’s ethical risks and how they can de controlled; development review, which will ensure the system’s operation is aligned with objectives and has accurately identified the harms and benefits; pilot review; where assumptions about the system’s performance will be tested in a controlled environment; and deployment review, which will ensure that processes for monitoring and review are adequate for controlling ethical risks over the lifetime of the system.
“UnionBank has been at the forefront of utilizing AI to provide the best experiences for our customers. But we didn’t want to stop there,” said UnionBank President and CEO Edwin Bautista. “We also want to stay true to our goal of promoting inclusive prosperity in the country, and one of the ways we are doing this is by sharing the knowledge that we have gained over the yearsin implementing emerging technologies, including AI, so that everyone can also benefit.”