CAN-ASC-6.2: Accessible and Equitable Artificial Intelligence Systems - Notice of Intent
This notice is to announce that Accessibility Standards Canada intends to develop a new accessibility standard. The purpose of this notice is to avoid duplication in standards development.
This notice is aimed at other standards development organizations and the public. It does the following:
- It defines the scope of the standard that we propose.
- It defines the need for the project.
- It states the time period during which the public can send us comments about the standard.
Name of standards development organization
Accessibility Standards Canada
320 boulevard Saint-Joseph
Gatineau QC K1A 0H3
The work program lists all of Accessibility Standards Canada’s in-progress and completed standards.
National Standard of Canada – Domestic
Standard development activity
03.100 – Company organization and management. Management systems
35.020 – Information technology in general
35.040 – Information coding
35.080 – Software
35.100 – Open systems interconnection
35.240 – Applications of information technology
Comment period start date
March 7, 2023
Comment period end date
March 28, 2023
The purpose of standard CAN-ASC-6.2 is to ensure the technical requirements for artificial intelligence systems are equity-based. This requires going beyond minimum technical specifications.
There are common areas where people with disabilities may face barriers related to artificial intelligence systems. These include:
1. Accessible Artificial Intelligence Systems
People with disabilities must be able to participate in the artificial intelligence economy. This standard will require that barriers be removed in various areas. This includes barriers to participating in, designing and developing, implementing, using, or improving artificial intelligence systems. It also includes barriers to evaluating (or being evaluated by) such systems. This includes tools related to:
- data visualization
- programming and coding
- marketed products and tools that include artificial intelligence processing
- consumer feedback systems and auditing.
2. Equitable Artificial Intelligence Systems
This standard will:
- require that artificial intelligence systems treat people with disabilities equitably
- cover all parts of the artificial intelligence system’s life cycle
- provide guidance on removing systemic bias against people with disabilities.
Bias in artificial intelligence systems can lead to the inequitable treatment of people with disabilities. It can occur in data, labelling, training, and algorithms. It can also occur when deploying or evaluating artificial intelligence systems.
The standard will provide guidance specific to people with disabilities in the following areas:
- the accuracy, including misinformation and disinformation
- bias, including statistical bias
- fairness in outcome and process
- safety and security
- oversight and protection of privacy
- protection from data misuse and abuse
- transparency and accountability
- individual agency (control) and informed consent
- choice in how artificial intelligence is applied
- human control and decision-making
- human judgment in decision-making
- cumulative harms.
3. Processes for Implementing Accessible and Equitable Artificial Intelligence Systems
Organizations need processes for putting accessible and equitable artificial intelligence systems in place. This standard will provide guidance on how to do this. These processes include:
- planning and justifying the use of artificial intelligence
- assessing impacts
- ensuring ethics oversight
- designing, developing, buying, and/or customizing artificial intelligence systems
- training users and operators
- providing access to alternative approaches.
This also includes processes related to:
- transparency, accountability, and consent
- feedback, complaints, redress (remedies), and appeals
- review, refinement, and termination.
Accessibility Standards Canada was created under the Accessible Canada Act. Its mandate is to prevent, identify, and remove barriers to accessibility in Canada for people with disabilities. To do this, it develops standards. These are based on the needs of people with disabilities. These needs are identified by consulting Canadians with disabilities. This includes the members of the Accessibility Standards Canada Board of Directors. Most Board members are people with disabilities. The Board has approved the development of this standard.
Accessibility Standards Canada standards are developed following the principle of “nothing without us.” This means that people with disabilities are involved in the development process:
- Its standards are developed using research that is led by people with disabilities or with lived experience.
- People with disabilities are members of the organization’s technical committees. These committees produce equity-based requirements for standards. These take into account the needs and perspectives of people with disabilities.
- The public review process for this standard will be accessible. This will allow even more people with disabilities to be part of the development process.
The use of artificial intelligence systems carries opportunities and risks. People with disabilities experience the extremes of both. They can face barriers and discrimination in every aspect of these systems. Also, the tools that allow participation in such systems are not accessible to many people with disabilities.
People with disabilities are very different from one other. However, their data is often lumped together, or they are viewed as being outside the norm (an outlier). They are also seen as a trivial minority. They are not understood, or they are misunderstood. Artificial intelligence systems may categorize them as “not optimal”. Existing artificial intelligence ethics protections fail to adequately consider the harms and barriers faced by people with disabilities.
The proposed standard is needed to:
- prevent harm that could be caused by artificial intelligence systems
- allow people with disabilities to have access to these systems and be treated equitably
- supplement and enhance current guidance and directives
- support the ethical implementation of artificial intelligence systems.
Posted to the Standards Council of Canada Centralized Notification System on March 1, 2023.