BRIEFING NOTE

by Maja Watanabe-Jovic

Summary:

• AI is a powerful tool. Its rapid growth requires the Canadian government to address and mitigate related national and personal security risks.
• AI should be regulated in a similar capacity within the public and private sector.
• Fostering connections between industry and government is necessary to ensure effective development and regulation of emerging technologies.
• It is critical that Canadian Defence treats data as an asset that needs protection. We must shift the focus from AI to also include data.
• Data reliability is at the centre of AI capabilities. Without accurate and effective data, we cannot have reliable AI models.

Context:

• Members of the Five Eyes Alliance (FVEY) are taking rapid steps to regulate and monitor the development of AI.
  • Australia, in particular, is moving towards greater regulation in this space as they have experienced data privacy breaches in the past concerning sensitive information (unrelated to AI).
  • They are working to create greater penalties for these data breaches when industries are not fulfilling their data safety and integrity responsibilities.
• AI is increasingly being implemented by governments in more complex projects, from simple business operations to defence capacities.
  • Associated risk levels must be regularly assessed across the spectrum of AI operations.
  • Activities such as AI document translation may operate at low risk while the usage of autonomous webcams can be considered to have greater security
  • It is essential to not focus solely on the safe conduct of AI but on the potential detrimental outcomes.
• AI is constantly and rapidly evolving. As a result, there is close competition among global actors to adopt the best practices and implementations.
  • Actors must adapt to this culture of rapid change.
• The development of a data culture is crucial to AI; we cannot have highly functioning AI without considering the quality of data.
• The hope for AI is that it will allow Canada to scale up its workforce as there continue to be recruitment challenges in the Canadian Armed Forces.
  • The use of AI can offset the scale of new members who lack training.
  • AI can fill in personnel shortages. It produces and optimizes the potential that work products possess.
• AI can greatly expedite the completion of documentation processes allowing for drafting and translation to be conducted simultaneously. The scale of the work being done is enormous.

 Considerations:

• To ensure accountability and compliance from relevant industries, there is legislation in place in Canada that provides regulation on oversight and accountability for AI integration, particularly for the private sector.
  • g., The Privacy Act and The Security of Information Act
• Contracting language is critical to effectively share information with allies and prevent misinformation.
• Governments continue to rely on industry for understanding the world of AI, however, there are many leading allies in this field that we should continue to look for lessons learned.
• Although it is important to regulate the use of AI, this may result in less innovation.
  • This can be detrimental, especially from a defence and security perspective. There is far too much competition among other players on the global stage to fall behind.
• Fostering connections with industries is of the utmost importance. It enables Canada to examine “good use” cases and try to implement them wherever possible.
• The government has some considerations in order to effectively communicate AI needs to industry.
  • Exercising care in vendor selection and engagement.
  • Clearly communicating government technological needs to industry partners.
  • There is a possibility for holding an “Industry Day” where vendors can follow up on proposals and the reverse - Government is the audience while industries showcase their capabilities.
  • This Industry Day must be an open call to all industry partners to be an equitable process.

 Implications for Canada:

• For the government to ensure that AI moderators respect the boundaries in the interest of the Department of National Defence (DND), they must remain transparent with the public about when and how their data is being used by their respective algorithms.
• DND aims to be AI-enabled by 2030
• Discussions between industry, government, and academia are crucial to scale up these opportunities for AI Integration.
• Canada must prepare to examine surveillance and awareness issues regarding the use of AI. Additionally, we must look at enabling AI to collect and share data in a way that is in the interest of national security.
  • There are heavy implications for personal security as well, given privacy legislation.
• DND has a responsibility to uphold an image of public safety and accountability. They must be careful with their biases and how they wish to use AI.

Conclusion:

• Transparency is key for the adoption and implementation of AI capabilities. Without it, the public has no control over how and when their data is used which can raise privacy concerns.
• Data sharing and data protection need to be embedded, not separate entities, for AI to be successful.
• We are rapidly approaching an age where AI can replace physical actors and complete tasks expeditiously. It will be greatly beneficial as a resource where we lack expertise and/or workforce.
• Canada must continue to implement and disclose best practices on data governance, both on the national level as well as among allies.
• The government has many considerations to make regarding what they want AI to do for them specifically, as well as for industry and vice versa.