Canada’s proposed AI law too vague, Big Tech executives say – National | 24CA News
Representatives from Big Tech corporations say a Liberal authorities invoice that might start regulating some synthetic intelligence methods is simply too obscure.
Amazon and Microsoft executives advised MPs on the House of Commons trade committee at the moment that Bill C-27 doesn’t differentiate sufficient between high- and low-risk AI methods.
The corporations mentioned abiding by the regulation as written could be expensive.
Nicole Foster, director of world synthetic intelligence and Canada public coverage for Amazon, mentioned utilizing the identical strategy for all purposes is “very impractical and could inadvertently stifle innovation.”
The use of AI by a peace officer is taken into account high-impact in all circumstances, she mentioned — even when an officer is utilizing auto-correct to fill out a ticket for a visitors violation.
“Laws and regulations must clearly differentiate between high-risk applications and those that pose little or no risk. This is a core principle we have to get right,” Foster mentioned.
“We should be very careful about imposing regulatory burdens on low-risk AI applications that can potentially provide much-needed productivity boosts to Canadian companies both big and small.”
Microsoft gave its personal instance of how the regulation doesn’t appear to distinguish primarily based on the extent of danger that individual AI methods introduce.
An AI system used to approve an individual’s mortgage and deal with delicate particulars about their funds could be thought-about the identical as one that’s used to optimize bundle supply routes utilizing public knowledge.
Industry Minister François-Philippe Champagne has been providing some details about amendments the federal government expects to place ahead to the invoice to make sure it’s up-to-date.
But despite that further element, corporations mentioned the definitions within the invoice are nonetheless too ambiguous.
Amanda Craig, senior director of public coverage at Microsoft’s workplace of accountable AI, mentioned not differentiating between the 2 would “spread thinly the time, money, talent and resources of Canadian businesses — and potentially mean finite resources are not sufficiently focused on the highest risk.”
Bill C-27 was tabled in 2022 to focus on what are described as “high-impact” AI methods.
But generative AI methods corresponding to ChatGPT, which might create textual content, photographs and movies, turned extensively obtainable to the general public solely after the invoice was first launched.
The Liberals now say they may amend the laws to introduce new guidelines, together with requiring corporations behind such methods to take steps to make sure the content material they create is identifiable as AI-generated.
Earlier this week, Yoshua Bengio, dubbed a “godfather” of AI, advised the identical committee that Ottawa ought to put a regulation in place instantly, even when that laws isn’t excellent.
Bengio, the scientific director at Mila, the Quebec AI Institute, mentioned a “superhuman” intelligence that’s as good as a human being might arrive as quickly as in just a few years.
Advanced methods might in the end be used for cyberattacks, he mentioned, and the regulation must get out forward of that danger.
AI already poses dangers. Deepfake movies, that are generated to make it seem like an actual particular person is doing or saying one thing that they by no means did, can be utilized to unfold disinformation, mentioned Bengio.
© 2024 The Canadian Press
