Microsoft details its support for Canada’s responsible AI framework

The Artificial Intelligence revolution is here, and a future powered by AI is becoming all the more certain.

However, with the advancement of AI, it is also important that innovation takes place responsibly and aligns with human values.

“With this exciting technological revolution, the role of human oversight cannot be overstated. AI is a tool harnessed by human ingenuity,” wrote Microsoft in its latest blog post. “Our ability to guide, govern, and ensure responsible deployment of AI is paramount. Regulation is crucial to helping ensure that, while we explore AI’s vast potential, we do so responsibly.”

According to the tech giant, Canada stands at the forefront of responsible AI governance.

From the recent launch of a Canadian code of conduct for advanced AI systems to the contribution to the G7 Hiroshima Process, Canada is demonstrating its commitment to responsible AI governance. The proposed Artificial Intelligence and Data Act (Bill C-27) is another milestone in establishing legislation that balances the risks and benefits of AI, while protecting consumer privacy.

The voluntary code focuses on 18 measures within six principles: accountability, safety, fairness and equity transparency, human oversight and monitoring, and validity and robustness.

Whether it is helping to fight cancer, speeding up building permits, or paired with Microsoft’s AI tech to preserve the Inuktitut language, prevent wildfires, and more AI-powered solutions are empowering Canadians.

Microsoft is committed to advancing Canada’s responsible AI framework. “Microsoft is deeply committed to this journey where Canada continues to lead, inspire, and thrive in the digital age,” wrote the tech giant.

However, it’s also worth noting that Microsoft’s own AI efforts haven’t been all that ‘responsible.’ A recent MSN article directed tourists visiting Canada’s capital to go to the Ottawa Food Bank and “consider going into it on an empty stomach.” The article was later pulled, and Microsoft said that the article wasn’t generated directly by AI, but through a combination of algorithmic techniques with human review.

Further, while Microsoft’s Bing AI image generator does a good enough job of blocking sensitive content, like child sexual exploitation, hate speech and graphic violence, its filters can easily be bypassed. For example, if you image search for queries related to “9/11,” “twin tower” and “terrorism,” the image generator would normally block it. But if you type in your query in an ambiguous way, like “scene from the cockpit of a plane flying towards two skyscrapers,” the image generator will give you a scene from the cockpit. As shared by 404Media, it was able to add family-friendly characters, like Mickey Mouse, Kirby, and SpongeBob, to such scenes. Read more about it here.

Elsewhere, Canada’s Bill C-27 has faced criticism. The bill’s proposed Artificial Intelligence and Data Act (AIDA) doesn’t include principles on how AI systems will be regulated, and they might also hinder private sector innovation. Read more here.

You can read Microsoft’s full report, titled “Meeting the opportunity and governing AI in Canada,” here.

Source: Microsoft