Artificial Intelligence is a moving target when it comes to standardization. AI is progressing at a breakneck speed and expanding its reach worldwide as it becomes broadly incorporated into products and services. Virtually any emerging technology requires standards to provide the foundation for safety, performance, and interoperability, but AI has additional opportunities and threats associated with its use calling immediately for standards to be set.
AI is a particularly complex technology since it can be used for information gathering, analysis, decision-making, and automation, and users are often unaware AI is driving it. For example, algorithms execute AI-based advertising on platforms like Facebook, targeting users according to their past and predicted-future behaviour. These systems do not automatically consider issues such as bias, discrimination, ethics, privacy, and human health and safety when they direct users to information or offer solutions.
The Canadian Mirror Committee to JTC 1/SC 42 Artificial Intelligence “AIMS” to help. The committee has successfully advanced a project proposal for the first conformity assessment standard for AI at ISO/IEC, having garnered unanimous international support in the ballot. The Artificial Intelligence Management System (AIMS) standard will enable organizations to show they have implemented and continually work on improving processes unique to the development or use of AI, such as bias, fairness, inclusiveness, safety, security, privacy, accountability, explicability, and transparency.
“Management system standards have been used in many different sectors to help innovation and technology develop through structured governance and appropriate risk management,” said Paul Cotton, Convenor of the Working Group 1. “One such example is the ISO 27001 for cybersecurity, which is currently used as part of the CyberSecure Program in Canada.”
“Having a standardized management system for AI is essential,” says Marta Janczarski, Standards Council of Canada (SCC) AI Sector Specialist and project editor for the AIMS standard. “Within our Innovation Initiative at SCC, we see firsthand incredible advances in technology and the wide-reaching applications possible for AI. But with that comes an innate responsibility to protect citizens from the dangers as well and providing standardized requirements and a means to demonstrate compliance is the best way to achieve this.”
AIMS will increase interoperability, harmonize requirements, increase trust in AI systems and ensure continual improvement through the standards development process that relies on balanced stakeholder representation, consensus, and public scrutiny as technology progresses. The full complement of a standard and associated conformity assessment program will give stakeholders confidence in an organization’s ability to consistently meet customer requirements, and any applicable statutory and regulatory requirements.
As Canada’s voice at the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC), SCC promotes national priorities by ensuring Canadian experts are at the development table shaping the international standards of tomorrow, like AIMS. This work also compliments initiatives with Innovation, Science and Economic Development to develop a pilot conformity assessment scheme for AI in Canada, as well as Canadian efforts under the Global Partnership on Artificial Intelligence.
AIMS is expected to be completed in 2023, acting in complement to developing regulations for AI, supporting further innovation and application of the technology.