There is global consensus in civil society, academia and industry that the introduction of artificial intelligence brings risks and harm. Addressing these concerns played only a minor role in Canada's national AI strategy. The federal government’s most significant response – the Artificial Intelligence and Data Act (AIDA) – is flawed and doesn’t address the present and noticeable impact of AI on our society.
Our research highlights key gaps in Canada’s approaches to AI governance. The first problem is that AIDA in its current form deals with government use. This is despite widespread use in the general public sector.
The Canadian Tracking Automated Governance (TAG) Registry lists 303 applications of AI in government agencies in Canada. The undeniable fact that AIDA as currently written won’t apply for presidency purposes implies that this laws is inconsistent with AI governance in other AI-leading countries and the US expressed interests of Government Employees.
That we all know so little about how the Canadian government is using AI is only one shortcoming we all know from a second report being released today. Our team also identified key gaps that span the last decade of AI governance in Canada. a part of Design AI project consists of research teams from Germany, the United Kingdom, Canada and France. Our report on AI in Canada documents a scarcity of critical discussion by all levels of presidency about AI and its risks, in addition to a failure to conduct public consultations.
Need for transparency
AIDA is Canada's first targeted attempt to control AI. The laws was appended to the top of Bill C-27 and is currently under review by the legislature Standing Committee on Industry and Technology. It was widely criticized for this don’t provide the protection Canadians need.
As Parliament debates AIDA, the federal government is accelerating the introduction of AI.
On April 7, the Prime Minister announced plans to spend $2.4 billion to extend AI adoption and use in Canada. Surprisingly, it was only 4 percent announced financing is devoted to the social impact of AI. These include vague existential risks, support for staff who could lose their jobs and a paltry amount for an upcoming AI and data commissioner.
More transparent use of AI by government and businesses, in addition to conducting meaningful consultations to strengthen oversight and accountability, would exhibit a real interest on the a part of the federal government to take public concerns seriously.
Our research shows a spot between the hopes and reality of AI that AIDA must close.
(Shutterstock)
AI registry
We developed this Canadian TAG Registry in collaboration with the UK-based company Public law project.
Numerous Organizations And state testing bodiesincluding Canada's Chief Information Officer Strategy Councilhave called for public registries for AI and automatic decision-making systems.
AI registries are already being created by a variety of cities, including Amsterdam, Helsinki, recent York And Nantes, France.
Our Canadian TAG registry is a start, but limited as a consequence of a scarcity of publicly available details about where and the way AI and automatic systems are getting used.
Document impacts
The argument for registers is predicated on the concept that policymakers and the general public have to see how government agencies and corporations are already using AI to develop effective oversight.
The management of this register – or an identical register – must be entrusted to an independent and well-resourced authority. This would facilitate a broader and more meaningful debate about whether, where and the way AI must be used and what sort of oversight we’d like.
There is a Extensive research work Documenting the ways wherein the usage of AI and automatic systems by governments and corporations has already caused harm. Previous research has also documented the burden faced by individuals, communities and testing sites Stop the usage of harmful AI practices as soon as they’re introduced.
The objectives of the Canadian TAG Registry are:
-
Advance discussion on the necessity for resourced, maintained and public registries of presidency and business use of AI and automatic decision systems (ADS);
-
enable a broader discussion about whether, where and the way AI and ADS must be used;
-
encourage more research and debate concerning the forms of systems used and their impacts;
-
exhibit the very limited information currently available about piloted or deployed systems.
Government agencies could be required to keep up and archive registers Dedicate sufficient resources to obviously record and communicate systems. In addition, government agencies would want to make procurement details and company processes more transparent, explain the intentions and uses of AI and automatic decision-making systems, and reply to citizen requests for information.
Proponents suggest that registries should include: Results of audits, details of the info sets and variables used and the way the system can be used.
(Shutterstock)
AI Governance in Canada
The federal government presented theirs Automated Decision Making Policy in 2019. This was intended to make government use of AI and algorithmic systems more transparent through mandatory impact assessments. At the time of writing, only 18 of them had been published.
The need for registration is justified A finding from our research. Our report documents remarkable standstills that AIDA didn’t address in the world Indigenous rights and data sovereignty in addition to a scarcity of input from the creative, cultural and environmental sectors.
Government policy has as a substitute narrowly focused on AI as economic and industrial policy. The consultations were largely theatrical, allowing AI adoption to proceed despite significant public concerns, particularly about facial recognition technologies.
Canadians' trust is suffering consequently. Canadians have certainly one of the bottom Level of trust in AIalthough Canada had certainly one of those first national AI strategies.
Even the federal government’s own procurement policy has largely ignored effective consideration of the social impact of AI. Instead, AI is seen as a cure for the service-oriented – or Deliverology – Canada’s public sector, and yet these changes were made with little public consultation.
AI has profound societal impacts, even when it is basically presented as an economic opportunity.
Withdrawal from AIDA
Our research critical reinforcements about Canada's recent efforts to control AI and points out two significant problems:
1) AIDA doesn’t apply to the usage of AI in the general public sector, although AI and automatic systems are widely used. This contradicts the concerns expressed by public sector employees. The Canadian Public Employees UnionThe Professional Institute for Public Employees And the Canadian Labor Congress have called for AIDA to be applied to government departments, agencies and Crown corporations.
2) AIDA was rushed and there have been none useful advice With the general public.
Given these limitations, AIDA not meets the needs of Canadians. Canadian laws also lags behind the regulatory approaches of other countries.
Examples of this are the The European Union’s latest AI law and the White House Supreme command And Guidancethat apply to AI uses by government institutions.
Canada stays behind the curve. The Prime Minister's current spending announcements won’t address the problems and challenges of regulating AI. AIDA must be separated from the remainder of Bill C-27 and sent back for public consultation, necessitating a recast.
