Guiding Principles

To successfully implement this Strategy by 2030, DND/CAF will be guided by the following principles:

  • We must put in place the data and technical enablers of AI. DND/CAF must implement the DND/CAF Data Strategy to support access to the high-quality, well-governed, well-architected, and appropriately classified data on which AI depends. DND/CAF must also create the secure and interoperable digital infrastructure required to support the application layer, including investments in the cloud infrastructure and security required to scale AI. This must be done with full consideration of Data Centric Security, an essential aspect of data and cyber security. DND/CAF must allocate financial and human resources to defining, implementing, and managing a secure space for AI application development.
  • We must embrace and actively manage change. We must overcome resistance to change and institutional skepticism by demonstrating the value proposition of AI. We must build on the lessons, results, and momentum of existing initiatives, learning from best practice and our own experience. Although leveraging AI is a process rather than an end, involving ongoing experimentation with emerging techniques and applications, we must also commit to fund projects to scale so that we reap the full benefits of our innovation.
  • We must recognize that AI is a means to solve a problem, not an end in itself. DND/CAF must maintain informed and realistic expectations of what AI can deliver, avoiding an inappropriate reliance on automated decision systems. We must create the conditions for success by ensuring that problems are appropriately defined and that the necessary data, technical and human foundations are present. We must approach AI wisely, choosing it where there is a reasonable expectation that its results will be superior to existing methods, with a clear understanding of its risks. Finally, we must view AI as a catalyst for a broader and more fundamental digital transformation of the business of defence, and an opportunity to imagine and design systems that are more transparent, equitable, and just.
  • We must deploy AI to augment, not replace, human action and decision making. Regardless of the tools, defence will always be a fundamentally human endeavour. We must ensure appropriate human involvement in AI systems, calibrated to their risk and impact. While this involvement may be minimal for low-risk applications, applications involving lethal force must always retain the human in the loop. Where possible, decisions and outcomes should be explainable and transparent, with appropriate accountability mechanisms in place.
  • We must not adopt AI without the processes demanded by AI. Advances in technology have compressed innovation cycles from decades to months. To succeed with AI, DND/CAF must be prepared to move at that pace. We must evolve our systems and processes to enable us to procure, develop, test, and field AI safely, securely and at the speed of relevance while working horizontally to achieve our AI goals. This will require that we incorporate agile project management, recognize software as a military capability as much as hardware, and integrate software agility and upgrades into our process to enable capability improvement. It will require that we solve critical process and capacity constraints for infrastructure and data. In short, we must be willing to accept fundamental disruption to benefit from AI.
  • We must calibrate our AI investments and ensure alignment to government priorities. DND/CAF must identify and prioritize areas for strategic investment in AI that will enable the development or amplification of key capabilities for priority missions, especially where AI can be a combat multiplier. Wherever possible, these should build on DND/CAF and Canada‚Äôs existing strengths in AI and defence to maximise the return on our investment and ensure sovereign AI foundations and capabilities. In calculating the return on investment, however, we must factor in the hidden costs of AI, including the costs of data and computation, and the impact on wider Canadian Government priorities and goals.

Page details

Date modified: