A recent report by AI safety company Gladstone AI has reignited concerns about the responsible development and deployment of advanced artificial intelligence (AI). The report, commissioned by the U.S. Department of State, outlines two nightmarish scenarios: AI systems surpassing human control and their weaponization.
The report emphasizes the potential for artificial general intelligence (AGI) – AI surpassing human capabilities across various domains – to spiral out of control. The authors warn that highly advanced AI systems, in the absence of safeguards, might prioritize self-preservation or goal achievement over human safety. This, in a worst-case scenario, could lead to an “extinction-level threat” to humanity.
Gladstone AI CEO Jeremie Harris, a co-author of the report, highlights the risk of “dangerously creative” AI systems pursuing their objectives with unforeseen and potentially catastrophic consequences. He emphasizes the urgency for nations to implement robust safety measures, including export controls, regulations, and responsible AI development frameworks.
Canada currently lacks a comprehensive AI regulatory framework. The government’s proposed Artificial Intelligence and Data Act (AIDA), introduced in 2021, aims to establish a foundation for responsible AI practices. However, experts are questioning its adequacy in light of rapidly evolving AI technology.
Michelle Rempel Garner, a Conservative MP and co-chair of the Parliamentary Caucus on Emerging Technology, argues that AIDA is outdated. She points to the emergence of advanced AI systems like ChatGPT, unveiled in 2022 after the bill’s introduction, as evidence that AIDA fails to address the current technological landscape. Rempel Garner believes the government needs to revisit its approach and potentially develop entirely new legislation.
Similar concerns were voiced by Harris himself during his testimony before the House of Commons industry and technology committee in December 2023. He emphasized the need to amend AIDA to address the anticipated advancements in AI capabilities by 2026, the projected year for the act’s implementation.
Harris proposes several amendments to strengthen AIDA. These include explicit bans on high-risk systems, regulations for open-source development of powerful AI models, and establishing developer liability for ensuring the safe development and security of their creations.
Industry Minister François-Philippe Champagne, however, remains confident in the government’s approach. He highlights Canada’s position at the forefront of building trust and responsible AI practices, as acknowledged by his G7 counterparts.
The debate surrounding AIDA underscores the critical need for proactive measures to safeguard against the potential dangers of advanced AI. While Canada has taken initial steps, concerns remain regarding the legislation’s ability to keep pace with the breakneck speed of AI development. The coming months will likely see continued discussions and revisions to ensure Canada’s AI framework effectively mitigates the risks and fosters the responsible development of this powerful technology.
