The use of artificial intelligence (AI) in employment decisions is controversial. While implementation of AI can make processes faster and easier, it also risks the introduction of biases and discrimination into hiring decisions.
As AI is a rapidly developing field, the legal framework with regard to employment has not yet caught up, leading to several current legislative developments both at the provincial and federal level.
The Risks of AI Use in Hiring Decisions
AI is trained on large datasets and makes decisions based on the information that is contained in the data. If the training data that is provided is not equally representative of individuals with varied demographic factors, then individuals may be unfairly discriminated against in hiring processes that rely on AI. This is the case when personal bias and discrimination have led to selection of a particular type of individual, which would then cause the AI to filter out types of applicants that do not match the characteristics of that type of individual.
Canada’s Position on AI Use in Hiring Decisions
The federal government has proposed new legislation, including the Artificial intelligence and Data Act (AIDA) in Bill C-27 to regulate the use of AI. Currently, it at the consideration in committee stage. In AIDA, AI systems are categorized based on their impact level, with “high-impact” systems being susceptible to causing harm to individuals and causing negative commercial effects due to bias.
The definition of “high-impact” was not included in the text of AIDA, but a suggested companion text to the amendments published by the Standing Committee on Industry and Technology as part of the discussions surrounding AIDA includes “matters relating to determinations in respect of employment, including recruitment, referral, hiring, remuneration, promotion, training, apprenticeship, transfer or termination” in the definition of “high-impact”. This is due to the fact that if unchecked, the use of previous data that may be drawn from discriminatory practices and can therefore perpetuate historical discrimination in the hiring process.
The text of Bill C-27 would require individuals using high-impact systems to create, monitor, and record measures to reduce the risk of harm or bias in the systems and disclose information about the use of the system and mitigation measures.
Ontario’s Position on AI Use in Hiring Decisions
Last year, the Ontario government proposed legislation entitled Bill 149, Working for Workers Four Act, 2023, which covers a number of amendments to existing legislation and includes provisions addressing the use of AI in hiring processes. As of February 12, it is on the Second Reading.
The proposed bill calls for amendments to the Employment Standards Act, 2000 to include public disclosure when AI is used in the hiring process, stating:
“8.4 (1) Every employer who advertises a publicly advertised job posting and who uses artificial intelligence to screen, assess or select applicants for the position shall include in the posting a statement disclosing the use of the artificial intelligence.”
However, section 8.4(2) states that the above provision “does not apply to a publicly advertised job posting that meets such criteria as may be prescribed.” Currently, the proscribed criteria are not provided, as they will be provided in regulations at a later time.
If enacted, this legislation would require employees to let the public know when AI is used in the hiring process, which is in line with the provinces objectives regarding AI. Ontario is currently creating guiding principles entitled Ontario’s Trustworthy Artificial Intelligence (AI) Framework that highlight transparency, trust, and benefit for all Ontarians. This provincial framework is intended to align with and complement the federal government’s currently developing position on AI.
Concerns with Ontario’s Proposed Legislation
Although Ontario’s proposed Bill 149 is a move in the right direction, several concerns have been raised by experts such as Dr. Teresa Scassa. In her submission in response to Bill 149, Dr. Teresa Scassa addresses the provisions applicable to hiring practices. Of particular concern are the lack of definition of “artificial intelligence” within Bill 149, the ambiguous nature of the required disclosure of the use of AI, the lack of recourse for individuals that apply for a job that utilizes AI, the limited scope, and the importing of bias into AI systems.
AI systems are often referred to as black boxes, as the decision-making processes are not visible to the users. Due to this phenomenon, it is possible that the provisions pertaining to the use of AI in hiring practices may end up being more thorough than originally proposed in order to sufficiently address the effects that the use of AI systems may have on the hiring process. As Bill 149 is still under consideration and early in development, changes may be made as it progresses through the Legislative Assembly of Ontario. Public hearings are scheduled to take place on February 12 and 13, 2024, and further discussion and amendments may provide some clarity into how the terms of the provision may be defined, what might be included under these provisions, and how it may look in practice.
Jeff is a lawyer in Toronto who works for a technology startup. Jeff is a frequent lecturer on employment law and is the author of an employment law textbook and various trade journal articles. Jeff is interested in Canadian business, technology and law, and this blog is his platform to share his views and tips in those areas.