Line of Effort 2: Change Management

The challenge

Leveraging AI broadly across the enterprise will require change and that change will need to be supported, monitored, and managed. The transformative and rapidly advancing nature of AI requires that DND/CAF actively fosters the expansion of AI across the enterprise by being more agile, innovative, inclusive, solution driven, and risk tolerant. These changes are needed not only to fully leverage AI, but they are also being demanded of the organization through its commitment to digital transformation, and by the rapid and often disruptive security, technical, industrial, and social changes currently taking place worldwide.

Our current structure, processes and incentives do not fully support the required change. Our focus and spending have been directed towards military hardware, and our structures and processes reflect that. Our processes limit our capacity to procure or collaborate with others to develop AI, hampering experimentation and innovation. Data and information are often disparate and unavailable, and business owners are unwilling to share it because of security concerns. Traditional styles of leadership have tended to vest authority at the top and reward conformity rather than innovation. This must change if we are to successfully implement AI.  

What we must do 

We must embrace digital transformation as foundational to adopting AI. Digital transformation and adopting AI will require our organization to accept a greater tolerance for risk and failure. AI research and development is high risk and often involves trial and error before success is achieved. We will need to embrace experimentation and appropriate risk-taking and provide protected environments in which this can be done safely. We will need to be tolerant of failing forward and early in the pursuit of experimentation and discovering what is possible.

We must learn to work horizontally. Working within commands or organizational silos will not yield the results we are seeking. Instead, we must learn to work collaboratively across the enterprise. Wherever possible, we should seek joint capacity and capability to reduce domain-specific stovepipes. Given the complexity of the technology, we must take an intersectional, multidisciplinary, and cross-functional approach to project design and problem solving. This horizontal, cross-functional approach will enable senior leadership to empower decision-makers at all levels with the authority required to identify and implement AI initiatives while embracing a diversity of perspectives as a strength rather than a weakness.

We must embrace disruption. AI is disruptive, and we must be prepared for the fact that adopting AI will bring changes—sometimes profound ones—to our structures and ways of working. We must embrace this. We must be willing to challenge orthodoxies and embrace novel methods of achieving objectives, making use of the innovation, diversity, agility, and excellence present within the Defence Team.

We must adapt continuously. In the past, changes to technology brought about episodic disruptions to culture, processes, and roles followed by periods of stasis. With the advent of AI and related technologies, innovation cycles have been dramatically compressed. We will need to be prepared to move to a state of continuous adaptation, alert for new developments and ready to integrate them into our corporate and battle space. In particular, the cost of experimentation, development, fielding and sustainment of AI will need to be built into defence capability development plans. DND/CAF will require greater nimbleness and agility in our supporting processes in both military and civilian branches, including the necessity of continual reviewing and updating of training and delegated authorities to maximize horizontal action of thought and purpose. Finally, our training system needs to fully incorporate AI whenever possible to enable quicker, more adaptive, individual, and collective force generation systems.

How we will do this

  1. Vest decision authorities for AI at the lowest appropriate level to encourage innovation. Although centres of excellence will be critical to AI success, local development is equally vital to encourage the development of solutions to meet specific needs. Consequently, leaders must be empowered with the authority to experiment and innovate within a horizontal, cross-functional environment while selecting and respecting the AI adoption processes, developed by the DCAIC, best suited to their specific circumstances.
  2. Identify and change incentives. We must identify the ways in which incentives such as promotions, recognition or career structures can be used effectively to encourage the kinds of behaviours required for AI adoption at scale. We will need to increase incentives to innovate and lower the costs and risks of failure by providing protected environments for early failure and adopting a no-blame approach.
  3. Include AI as a defence capability enabler that requires funding. If it is to realize its AI aspirations, DND/CAF must commit to fund the technical, digital, and data enablers and the research, engagement, and staff AI requires. These enablers must be costed and those costs built into program and project planning and development from the inception. New programs, projects, and initiatives will be required to assess the potential implementation of AI, and existing capabilities will be assessed to determine the need to integrate AI enhancements into legacy systems.
  4. Identify, prioritize, and address key impediments to responsibly procuring, developing, testing, validating and certifying, fielding and decommissioning AI. In addition to funding and supporting AI enablers, we must also ensure that key AI governance bodies have the authority and resources to minimize policy and process barriers to development, implementation, and adoption.

Policies and Standards Chatbot

AI capability needed: AI chatbot which can provide answers to questions on military dress policy

AI techniques used: Natural language processing

Value proposition: Improved policy comprehension and compliance among CAF members

Policy documents contain vital information, but they can be long and difficult to navigate for personnel looking for answers to specific questions. In response to this challenge, the Digital Transformation Office’s Data Science team has produced a working prototype which could help CAF members find answers to questions about military dress policy. 

Built from scratch by the Data Science team using industry-standard open-source technologies, the chatbot enables users to input questions in natural language and receive references to the passage the system believes contains an answer to their question, with links to relevant policies. The prototype also contains features allowing users to provide input on the relevance of the response to improve the accuracy of the tool and send feedback to the team on their experience. According to its developers, the chatbot is over 90 percent accurate in its responses to questions. This chatbot has only been deployed in test environments as a proof of concept. This class of AI tool holds promise to answer questions on other types of policies.

Page details

Date modified: