The use of artificial intelligence (AI) is now commonplace throughout society. The adoption of AI is driven by its utility and the improvements in efficiency it creates. Every day, most of us rely on AI for tasks like autocompleting our text messages, navigating our route to a new location, and recommending what movie to watch next. Beyond these common uses of AI, there are also uses that regulators are beginning to identify as areas where there may be a higher risk. According to the Digital Strategy website from the European Commission, these higher-risk areas can include uses of AI in employment, financial, law enforcement, and healthcare settings, as well as other areas where outcomes can have a significant impact to individuals and society . As the adoption of AI grows there is increasing recognition that enabling trustworthy AI is important, with 85% of consumers and 75% of executives now recognizing its importance . Establishing principles, such as IBM’s principles for trust and transparency, are important for guiding the development and use of trustworthy AI [2, 3]. Central to putting these principles into practice is establishing the appropriate governance mechanisms for AI systems.
AI governance will require an agile approach
AI governance is an organization’s act of governing, through its corporate instructions, staff, processes, and systems to direct, evaluate, monitor, and take corrective action throughout the AI lifecycle, to monitor whether the AI system is operating as the organization intends, as its stakeholders expect, and as required by relevant regulation. We expect regulations focused on AI systems to evolve rapidly and operators and developers of AI systems will need to adjust quickly as policy initiatives and new regulations are passed. Agile approaches, first championed in a software development context, are based on values that include collaboration and responding to rapid change . Agile governance approaches are now being used by governments around the world to respond quickly as technology advances and help enable innovation in emerging technology areas such as blockchain, autonomous vehicles, and AI . The use of an agile approach in AI governance can help AI adopters to identify whether changes in governance and regulatory requirements are integrated appropriately and in a timely manner.
Integrating RegTech into broader AI governance process
An agile approach to AI governance can benefit from the use of RegTech to meet the expected regulatory requirements for AI systems. As defined in the “Regulatory Technology for the 21st Century” World Economic Forum white paper, RegTech is “the application of various new technological solutions that assist highly regulated industry stakeholders, including regulators, in setting, effectuating and meeting regulatory governance, reporting, compliance and risk management obligations” . Examples of RegTech include chatbots that can advise on regulatory questions, cloud-based platforms for regulatory and compliance data management, and computer code that helps enable more automated processing of data relating to regulations . These RegTech solutions can operate as part of a wider AI governance process and be integrated as components of broader AI governance mechanisms that can include non-tech components such as an advisory board, use case reviews, and feedback mechanisms . Integrating into existing processes can be helped by strong stakeholder buy-in and beginning with basics such as a clear definition of AI, internal policies, and clarity on current legal requirements.
Case studies on OpenPages: Using RegTech for AI governance
IBM OpenPages with Watson is a RegTech solution that can help adopters to navigate an environment with rapidly changing regulatory and compliance demands . IBM has helped clients such as Citi, Aviva, General Motors, and SCOR SE to leverage this RegTech to help address governance requirements, mitigate risks, and mangage compliance [8, 9, 10, 11, 12]. IBM is also using IBM OpenPages with Watson as a foundational RegTech component in its internal end-to-end AI governance process. IBM OpenPages with Watson can help enable the collection of compliance data on AI systems to help evaluate compliance against corporate policy and regulatory requirements. The use of RegTech for AI governance from the early outset of regulatory requirements for AI systems can help enable the creation of a centralized regulatory library to facilitate collection of data and tracking where data would otherwise exist in silos across the business. By leveraging a centralized RegTech solution, the business can also potentially benefit from efficiencies in the processes and resources enabling these solutions.
Looking forward: RegTech expected to play a central role in AI governance practices
We predict RegTech will play a central role in AI governance practices in 2022 and beyond. We expect RegTech solutions will continue to adapt to meet the needs of companies who will be impacted by new regulations, standards, and AI governance requirements. AI is also likely to drive unique requirements for specific RegTech functionality relating to bias assessments (which could include specific metrics like disparate impact ratio), automated evidence to monitor for drift in AI models, and other functionality relating to the transparency and explainability of AI systems.