MPAC AI Roundtable

Our vision is to harness the power of responsible AI to improve public services, enhance decision-making, and deliver better outcomes for communities across Ontario. The AI Roundtable enables us to:

• Hold close, focused discussions and explore a specific topic with all participants on equal footing.
• Confront issues collaboratively rather than focusing on individuals, promoting open dialogue and problem-solving.
• Facilitate diverse input and break down hierarchical barriers that often silence less powerful voices in meetings
• Promote deep, free-flowing discussion, sharing of ideas, and decision-making based on a variety of perspectives.
• Enhance collaboration, networking, and team building by encouraging equal participation and engagement among members.
• Drive innovation. Insights from the round table can inspire new ideas and strategies, helping MPAC remain competitive and innovative in its use of AI.

By joining forces with like-minded professionals, we are creating tools and frameworks that will guide AI adoption across government and public institutions. 

The expertise and insights of our Roundtable members are invaluable to this initiative. By joining, they have the opportunity to:

Shape the future

Contribute to developing innovative AI applications to enhance public services and improve societal outcomes.

Collaborate with leaders

Engage in meaningful discussions with fellow thought leaders and experts across various fields.

Influence policy

Help shape governance models and ethical guidelines that will steer AI adoption in a responsible and sustainable direction.

Warren Ali
Director, Industry Development, Vector Institute

Warren is Director of Industry Development at the Vector Institute. He is a seasoned professional with extensive experience in industry innovation and management consulting. Warren began a career in the financial sector as a Senior Analyst within the Walmart - SmartCentres JV Property Portfolio at SmartCentres REIT. Warren holds a Master of Business Administration (MBA) in Finance and International Business from the Schulich School of Business at York University, obtained between 1997 and 1999, and completed coursework at the Canadian Securities Institute in 1998.

Chad Cogar
Vice President of AI, Creative Destruction Lab

Chad leads AI initiatives at the Creative Destruction Lab, a global program for science-based ventures operating 11 locations in 6 countries. He is the technical lead for the CDL's "Putting AI to Work Program," which empowers organizations to boost productivity through the adoption of AI.

With over a decade of experience in AI product strategy and implementation, Chad specializes in creating AI systems that drive measurable business impact. He previously led product for the machine learning teams at Kindred (sold to Ocado for $260m), and then led large AI transformation.

Mark Daley
Chief AI Officer, Western University

Mark is the Chief AI Officer at Western University and a full professor in the Department of Computer Science with cross-appointments in five other departments, The Brain and Mind Institute, The Rotman Institute of Philosophy, and The Western Institute for Neuroscience. He is also a faculty affiliate of Toronto's Vector Institute for Artificial Intelligence.

Mark has previously served as the Vice-President (Research) at the Canadian Institute for Advanced Research (CIFAR), and Chief Digital Information Officer, Special Advisor to the President, and Associate Vice-President (Research) at Western. Mark is the past chair of Compute Ontario and serves on a number of other boards.

Nihar Dalmia
Partner, Omnia AI, Deloitte

Nihar is a Partner in Omnia AI, Deloitte’s artificial intelligence (AI) practice. He leads the Insights & Engagement and Automation teams made up of Data Scientists, AI Engineers, Designers, Data Visualization experts, and Automation Engineers. Nihar specializes in driving value from AI and Data solutions, from strategy through implementation. He has worked with organizations across Asia, North America, and South America to deliver data-driven insights.

Nihar is passionate about solving complex business problems using exponential technologies to drive both economic and social impact. He has a Master of Business Administration degree from the University of Oxford and a Master of Science degree from Massachusetts Institute of Technology.

Jesslyn Dymond
Director, Data Ethics, TELUS

Jesslyn Dymond is the Director of Data Ethics at TELUS leading the approach to responsible data-driven innovation, drawing on a background of privacy and information management expertise. She is part of TELUS' Data & Trust Office, a team devoted to ensuring that data handling practices are responsible and respectful.

Jesslyn is a CIPP/C and has completed a Masters of Information from the University of Toronto’s iSchool. She is recognized for her leadership in data enablement and as an advocate for creative solutions to building customer trust with emerging technology.

Dawn Hall
Senior Advisor, Responsible Data and AI, Treasury Board of Canada Secretariat

Dawn is part of a team that sets policy and provides guidance for the responsible use of AI and ethical use of data across the federal government. This includes the stewardship of policy instruments such as the Directive on Automated Decision-Making, the Algorithmic Impact Assessment Tool (AIA) and the Guide on the use of Generative AI. Dawn also has experience in data strategy and governance, including work on the data strategy for the federal public service. Prior to her current role, Dawn worked in science communication at science centres and museums, having previously obtained a PhD and postdoctoral training in plant biochemistry.

Simon Hodgett
Partner, Technology, Osler

Simon’s practice focuses on technology related commercial contracting and advice. He advises enterprises whose businesses rely on technology and complex services. He also advises technology suppliers ranging from large established software providers to early stage technology companies.

He leads a team providing innovative legal services models to the firm’s high growth technology clients pursuing a wide variety of cutting edge technology and business models, including FinTech, artificial intelligence, data analytics and eHealth. Simon has been lead counsel on projects across a broad range of industry verticals, including the technology, banking, pension, investment, healthcare, energy, telecommunication and retail sectors. He also advises on selling technology to governments and large enterprises.

Sam Ip
Partner, Technology, Osler

Sam is a member of the Technology Group that regularly works with artificial intelligence companies or organizations making use of artificial intelligence technologies. This includes helping clients understand evolving global standards and regulations, assessing potential risks and liabilities associated with this emerging technology, and identifying processes for integrating artificial intelligence risk management strategies into organizational activities.

Fatima Khamitova
Director, Startups and Scaleups, Vector Institute

Fatima has over 10 years of experience in AI through consulting, marketing, and startups. Currently, she leads Startups and Scaleups teams at Vector Institute - leading global AI research facility located in Toronto, Canada. She advised a wide range of clients in retail, entertainment, and financial institutions on topics of AI, digital transformation, and analytics. Fatima is a strategic thinker who transforms customer experience using AI-based tools. She developed and implemented roadmaps for her clients, spanning a full range of services from marketing and operational tactics to implementing sound financial decisions to set her clients up for success. Fatima also speaks at universities and schools, client events, and panels.

Nicole McNeill
President, MPAC

As President of MPAC, Nicole is responsible for leading the organization as it continues to provide valuable service to the province of Ontario and advance MPAC’s global leadership in property assessment.

Nicole has 25 years of experience in the non-profit and private sectors and a passion for building diverse, high performing, engaged teams. Nicole currently serves as Chair of the Board of Elexicon Energy Corporation. She previously chaired an IT Advisory Panel and served on the Board of the Canadian Urban Institute. She has also been a mentor in the Top 200 Management Program at Toronto Metropolitan University (formerly Ryerson University) since its inception. She serves as a mentor to MPAC’s annual Continuing Academic Excellence President’s Award recipient.

Sonia Sennik
Chief Executive Officer, Creative Destruction Lab

Sonia is CEO of Creative Destruction Lab, a global program for science-based ventures operating 11 locations in 6 countries. She also led the CDL Rapid Screening Consortium during COVID-19. With a decade at Hatch managing large projects, she founded the Sonia Sennik Resilience Fund at McMaster University, the largest multi-donor endowment in its history. Named Canada’s Top 40 Under 40, she serves on the board of Futurpreneur Canada, and was former coach of the Senior Women’s Rugby Ontario team.

Objectives

The objectives of this roundtable are to:

1. Advise MPAC on how to implement artificial intelligence (AI) to innovate in public sector initiatives while reasonably protecting against foreseeable risks.
2. Share and exchange information and best practices regarding how public sector organizations are currently and should be implementing AI, as well as the challenges and opportunities with respect to its implementation.
3. Share and exchange information about evolving international trends, legal, policy and technical standards on the use of AI in public sector organizations.
4. Bring together industry and public sector experiences in AI implementation.
5. Establish MPAC as a leader in the use of AI in the public sector.

Membership

The roundtable will be comprised of a maximum of 12 of members consisting primarily of (i) academic experts, (ii) legal experts; (iii) technical experts; (iv) entrepreneurs and members of the business community; (v) non-profit organizations; (vi) MPAC’s internal stakeholders.

Deliverables

For a minimum of one year, MPAC’s AI Roundtable will strive to achieve the following deliverables:

1. Create a roadmap outlining MPAC’s goals for its AI initiative and a roadmap for achieving those goals.
2. Develop a range of potential use cases for MPAC to innovate using AI.
3. Develop a framework, including a risk management and accountability framework specific to MPAC’s proposed use cases.
4. Formulate a set of guidelines to steer public sector organizations in adopting AI responsibly and effectively, tailored to the sector’s needs.

Key success indicators

Should MPAC’s AI Roundtable succeed in fulfilling its objectives during its term, the following success indicators will hopefully begin to appear in whole or in part:

1. MPAC will have gained a better understanding of what responsible and innovative use of AI means for a public sector organization.
2. MPAC will be ready to confidently use AI to drive innovations in its services.
3. MPAC will be able to demonstrate its ability to use AI accountably to regulators and to the public.
4. MPAC will be able to use AI in its services with the trust of consumers and the public.
5. The policies, discussion papers and other documentation developed by the roundtable will be used as a templates for other public sector organization looking to use AI.

Meeting 1

Attendees: Nicole McNeill, Sujit Jagdev, Scott Milne, Julia Lipman, Jeffrey Ma, Tammy Wells-Garrett, Stratos Kaloutas, Warren Ali, Fatima Khamitova, Nihar Dalmia, Dawn Hall, Chad Cogar, Sam Ip, Simon Hodgett. 

Meeting notes summary:

• AI Register for the Federal Public Service: Dawn Hall presented on the newly published federal AI register, explaining its purpose, development process, and scope, with active discussion and feedback from the roundtable on transparency, governance, and sectoral practices. 
  • Register Purpose and Scope:  AI registers boost transparency and accountability as they provide standardized, searchable databases that document information about AI tools used by government organizations. Certain AI tools may not be included in a register, such as those relating to national security. 
  • Development and Future Updates: The Government of Canada register was created as a minimum viable product by collating existing information and having it validated by the relevant government departments. Going forward, there will be an iterative process between gathering feedback on how the content and features of the register can be improved, and using such feedback to improve the register through regular updates.  
  • Broader impacts of the register: The roundtable discussed how this register may be used by private sector organizations as an example of how their own AI registers should look. Additionally, the use case descriptions may be helpful for companies to reference as they develop and provide the government with complementary AI products and services in the future. 
     
• Discussion on Canada’s Cloud Foundation: Sujit Jagdev led a discussion on establishing a national-level open-source foundation that functions as a hub for open-source projects in Canada. 
  • Foundation Concept and Rationale: The foundation would position itself as a non-profit, developer-centric platform for incubating open-source projects and providing public-interest technology services.   
  • Sector-Specific Initiatives: The foundation could potentially be organized by sectors, such as agriculture and healthcare, to facilitate targeted collaboration and data sharing to make participation less daunting and more impactful for the stakeholders involved. 
  • Governance and Funding Models: The roundtable also discussed governance structures, referencing the Apache Software Foundation, the need for a central manager, and potential funding models, including government support and organic growth through academic and industry collaboration. Simon Hodgett proposed a three-tier structure, with the foundation at top, followed by sector-specific groups, and lastly project-based initiatives, suggesting collaboration with existing organizations and academic researchers to maximize impact. 
  • Canadian Sovereign Cloud and Technical Architecture: Sujit Jagdev also presented a vision for a Canadian sovereign cloud, detailing technical components and explaining the strategic advantages of building a Canadian sovereign cloud, citing geographic stability, talent availability, and the need to reduce dependency on foreign hyperscalers like AWS and Microsoft. 
  • Open-Source Software and target audience: The proposed cloud architecture would leverage open-source technologies and offer Canadian companies, governments, and startups a domestic and more secure alternative to existing cloud services providers. 
     
• Roundtable Retrospective and Forward Planning: Nicole McNeill facilitated a retrospective on the roundtable's first year and discussed key takeaways and suggestions for future meetings. 
  • MPAC Key Takeaways and Impact: Nicole highlighted the development of selection criteria for choosing internally-driven AI projects, governance practices, and the importance of scalable, repeatable projects, while others noted the value of cross-sectoral insights and practical case studies. 
  • Collaboration and Knowledge Sharing: The roundtable emphasized the benefits of sharing experiences, challenges, and lessons learned, with suggestions to increase exchanges among organizations. 
  • Implementation and Lessons Learned: The roundtable also discussed having more examples and lessons on AI implementation, particularly around reasoning, privacy, interpretability, and agentic models to inform future scaling and productivity improvements.  
  • Upcoming Initiatives and Outreach: Nicole announced plans for an MPAC roadshow to share its story and lessons with municipalities. 

Wrap-up and next steps: 

• Advance Distribution of Discussion Topics: Going forward, the group will also try to share presentation materials in advance and dedicate time at each meeting for participants to share recent learnings. 
• Social Media Collaboration on AI Register: Coordinate with Osler to create and share a social media post or content piece highlighting the Government of Canada AI register and its relevance for technology clients.   

Meeting 1

Attendees: Mark Daley (virtual), Jesslyn Dymond, Nihar Dalmia, Chad Cogar, Dawn Hall, Fatima Khamitova, Nicole McNeill, Sam Ip, Simon Hodgett

The Municipal Property Assessment Corporation (MPAC) provides property values, insights and services to taxpayers, municipalities, governments and businesses in Ontario. MPAC maintains a database of information for over five million properties in Ontario with six trillion dollars of value. The size of this database is equal to more than a third of the size of Netflix’s database. 

MPAC is a public sector organization that operates with private sector DNA. While many public sector organizations are ordinarily risk-adverse, MPAC has been a leader in the adoption of new technologies. It was one of the first public sector organizations to fully adopt cloud-based services through Amazon Web Services (AWS) in 2015. MPAC seeks to bring that same leadership and innovative spirit to the adoption of artificial intelligence. That leadership has already begun through the development of “The Pitch” which asked MPAC employees to submit ideas about how AI could help them do their jobs better daily. Over 100 unique submissions were received. AI is understandably creating some nervousness in the working world as employees wonder whether their livelihoods will disappear, however, MPAC has sought and received full organizational buy in which has allowed it to begin this AI journey in an open and thoughtful way. 

In commencing this journey, MPAC has retained Simon Hodgett and Sam Ip of Osler, Hoskin & Harcourt LLP for their experience in the field of AI. They have observed a number of things related to the emergence of AI: 

1. Canada is consistently ranked highly regarding educational, research and development capabilities available for AI. However, it lags on success of implementation and adoption. Practical solutions and use cases are needed for AI to take the next step towards implementation in Canada. 
2. The public sector faces challenges pursuing the adoption of AI. Conversely the public sector is a resource constrained environment where the benefits of AI will likely be significant.  

During the morning of Friday January 17, 2025, MPAC invited industry leaders to Osler’s Toronto offices for the first of four roundtable meetings on the topic of artificial intelligence and its adoption at MPAC modeling approaches to adoption in the public sector. The roundtable sought to begin a dialogue on successes, challenges and the next frontier in this journey. During this discussion, certain themes and topics emerged which are set out in general terms below. 

The topics of discuss in this first session were as follows: 

• Automation of Routine Tasks: Generative AI has been adopted in government to automate routine tasks in the workplace. For example, as Canada is a country with two official languages, there is always a need for translation services. AI has been successfully deployed to support the work of translators and allow them to complete their work more efficiently. However, from a federal government perspective, the Official Languages Act mandates that government services be provided in both languages at the same level of quality. While AI is useful in increasing productivity, there is still a need to keep a human in the loop to ensure the work of AI meets the quality standards mandated by the Act. In addition, AI has been successfully deployed as public facing in the form of chatbots which can get consumers the information they need without the need to speak to a human. However, further work remains in this regard to ensure that chatbots can undertake and resolve complex inquires with or without human supervision. 
   
• Standards: Responsible AI is an essential part of the adoption journey. There is value in the development of a “Responsible AI Framework” which can ensure that as advancements continue in AI technology, fairness, privacy and overall well-being of individuals and society is prioritized. While Canada has seen some convergence with global standards through ISO, more work remains towards implementation of those standards. The implementation of standards can ensure appropriate transparency and trust of an organization’s workforce through the AI adoption process, and it can further ensure that individuals are educated on the ways in which AI may change and effect the direction of their jobs and the organization writ large. Approaching AI adoption organizationally from the bottom up, prioritizing space for experimentation and discussion, will ensure greater and more sustained success. 
   
• Change Management Plan: The AI journey requires appropriate preparation through a change management plan. Organizations that have utilized such plans have found that the significant and complex organizational changes that AI cause are integrated more successfully and individuals within the organization are more prepared for the changes caused by AI leading to less disruption or apprehension. Organization that didn’t have such a plan in place found that use cases were less impactful and more compartmentalized. 
   
• Procurement: In order to improve on AI procurement, it is vital to start with a desired business outcome prior to with AI rather than simply setting out to adopt AI.  Generally, AI strategy must be linked with the business strategy. 

With three more roundtables remaining in 2025, MPAC seeks to have a fulsome guide to the adoption of AI in the public sector. The next session will build on issues discussed in the first roundtables. 

Meeting 2

Attendees: Jesslyn Dymond, Nihar Dalmia, Chad Cogar, Dawn Hall, Tony Gaffney, Warren Ali, Nicole McNeill, Sam Ip, Simon Hodgett

Regrets: Mark Daley, Fatima Khamitova, Sonia Sennik, Humera Malik, Iliana Oris Valiente

Meeting Notes Summary:

• AI Strategy and Adoption: Tony Gaffney discussed the importance of adopting AI to stay competitive and sharing ‘good practices’ within the industry. The biggest risk is missing opportunity, and instead organizations should be instilling AI literacy, education, awareness, trust for employees and customers.
  • Good practices have 5 key elements: Clear aspirations, not looking at AI as a separate thing, the need for talent to compute data, efficiency in the use of AI, and collaboration
     
• Start-Ups: The question was posed on how an organization like MPAC can help find Canadian start-ups to help build, invest in, and bring into the fold. This will potentially be brought back to the next roundtable meeting.  
   
• The Pitch: Nicole McNeill and Scott Milne explained the process of gathering AI ideas from employees for MPAC’s ‘The Pitch’, resulting in over 130 submissions. They emphasized the importance of involving employees in the AI journey to alleviate fears of job loss and encourage innovation.
  • Design Thinking Workshop: Scott Milne described the design thinking workshop facilitated by Osler, which helped employees refine their AI ideas. The workshop allowed participants to focus on the problem and collaborate on potential solutions, fostering a culture of innovation.
  • AI Use Cases: Scott Milne presented the top nine AI use cases selected by employees, including AI knowledge base for valuation assessors, AI solution for legal interpretation requests, and automated building structure creation from plans. These use cases aim to improve efficiency and productivity within the organization.
  • Criteria for Evaluating AI Projects: Scott Milne shared the criteria for evaluating AI projects, including business value, impact, feasibility, risk assessment, and process characteristics. The criteria aim to ensure that selected projects align with strategic goals, provide measurable benefits, and have a clear project owner and executive sponsor.
     
• Federal AI Strategy: Dawn Hall briefly discussed the recently released AI strategy for the Federal public service, which aims to accelerate AI adoption within the government. The strategy focuses on central AI capacity, talent development, and building trust through transparency and engagement.

Meeting 3

Attendees: Nicole McNeill, Jesslyn Dymond, Sam Ip, Fatima Khamitova, Nihar Dalmia, Simon Hodgett, Jeffery Ma, Scott Milne, Tammy Wells-Garrett, Stratos Kaloutas, Julia Lipman, and Dawn Hall (virtually).

Meeting Notes Summary:

• Federal Government Initiatives on AI: Dawn Hall gave a presentation on the Federal Government Initiatives with AI, including the G7 GOV AI Grand Challenge - Rapid Solutions Labs.
  • AI-use in the Federal Government: AI has been used in the federal public sector for many years. However, there is no consistent tool across all government departments, and the federal public sector is seeing variability between different departments. The Federal government is seeing an increasing use of generative AI. AI-uses in government are constantly evolving and there are continued discussions on what tools will be broadly available.
    • Provided examples of AI-uses across the federal government:
      • Scientific Uses: Fisheries and Oceans Canada use AI to detect marine mammals using satellite imagery; Public Health Agency uses AI for disease prediction and detection; Canada Food Inspection Agency uses AI to improve quality control of crops.
      • National Research Council: Collaborating with Indigenous experts to develop AI to contribute to the revitalization of Indigenous languages.  
      • Service Delivery: For example, Immigration and Refugees and Citizenship Canada has many different AI-uses for applications including to support eligibility determinations; Social Development Canada uses AI to support the delivery of employment insurance. 
      • RCMP: The RCMP uses AI to assist with analysis on cases related to human trafficking and child exploitation. This also benefits employees as it eliminates the need for employees to be repetitively looking at challenging and harmful images.
      • Transport Canada: Using AI to evaluate risk of inbound cargo.
      • Public Services and Procurement Canada: Looking to use AI to assist with backlogs and pay systems.
      • Agriculture and Agri-food Canada: Using generative AI (AgPal) which pulls information from a variety of agriculture programs to help the public understand what grants are available and the applicable regulations.
      • Shared Services Canada: Developing a generative AI chatbot. 
         
    • Highlighted key partners in the federal government AI ecosystem:
      • Treasury Board of Canada Secretariat: Sets rules for the responsible use of AI in government.
      • Shared Services Canada: Looks at IT procurement for the Government of Canada.
      • Service Delivery Departments: Looking to implement AI to improve how clients are served.
      • Public Safety: Focused on security and privacy of government data. 
      • Statistics Canada: Holds significant amounts of data.
      • Innovation, Science and Economic Development Canada: Responsible for the digital economy and private sector regulation of AI.
         
  • Discussion on Minister Solomon’s Priorities – Minister of AI and Digital Innovation
    • The Minister is supported by ISED, with a mandate focused on economic leadership. There are four pillars to the Minister’s vision:
      • (1) Scaling AI: Scaling Canadian AI industry and championing Canadian innovation leaders.
      • (2) Adopting AI: Encouraging businesses to adopt AI to drive productivity and growth.
      • (3) Trust: Ensuring Canadians can trust the responsible use of AI (including privacy).
      • (4) Sovereignty: Building a more secure Digital Canada (through sovereign data centers and a secure cloud).
        • There is no firm definition of sovereign AI yet. It’s an evolving discussion.
           
  • AI Source List: The AI source list is a procurement vehicle to assist the federal government in purchasing AI services through pre-qualified vendors. However, there still has to be a procurement process and there has been limited uptake by federal government departments as the procurement can be more complicated.
    • There are currently 140 vendors on it with ongoing quarterly reviews. Vendors are selected based on three criteria: (1) experience; (2) qualification of team; (3) experience using an ethics framework for responsible AI. Vendors can be qualified at one of three levels, differentiated based on experience and the upper limit of work they can do (at a dollar amount).
  • G7 GovAI Grand Challenge – Rapid Solutions Lab: At G7 Leaders’ Summit, the G7 gave a statement on AI for Prosperity, which included a commitment related to adopting AI in the public sector. The government is now launching a G7 challenge with rapid solution labs that will operate like a hackathon. There will be a series of problem statements to see what participants can develop over the course of 2 weeks. To facilitate this, a G7 AI Network (GAIN) is also being established.
    • Currently developing problem statements and judging criteria.
    • The rapid solutions labs will take place in November 2025, with registration occurring in mid to late October 2025. There is currently no ongoing promotions – it was announced in June and is anticipated to have a greater presence in the fall as it is still being solidified.
       
• Provincial & Federal AI Initiatives: Discussion on how to align with or support federal AI leadership and how provinces can benefit from the Federal Government’s endeavors and leadership.
  • There is significant fragmentation between the provinces using AI.
  • The Federal Government is connecting regularly with provinces and are having conversations with provinces on what they are doing and how they are looking to adapt.
  • Alberta, Quebec, and B.C. are supporting AI more than Ontario and are both doing interesting work with it.
     
• Updates:
  • Jesslyn Dymond shared experience on participating in Mila’s indigenous AI conference with indigenous AI startups, researchers, and projects. Mila also delivers an Indigenous Pathfinders in AI program which is a six-week training program for indigenous students to learn machine learning. The program had incredible pitches.
  • Fatima Khamitova explained that Vector recently concluded a big project for sponsors on Agentic AI. The project had 32 teams participated and wrapped up today. There were many different AI-use cases employed. Also noted that a recent McKinsey report on Agentic AI provided that 8 out of 10 organizations that adopted it have not seen any financial gain from it.
  • Jesslyn Dymond shared that TELUS published a responsible AI report concerning the views of Canadians. The research involved 5000 Canadians and found that 1% of Canadians trust AI without human oversight.
     
  • Summary of Progress on AI and the Pitch at MPAC:
    • Scott Milne presented on the progress of the AI and the Pitch at MPAC.
    • Two of the three selected pitches are starting to move forward. One solution is focused on bringing together training documents and processes with a better search assistant/chatbot to help valuation staff. The other solution plans to streamline the Legislation Interpretation Request (LIR) process. The third solution is an AI automated system for reading building structure architectural diagrams from plans. This solution is far reaching and the most complicated because there is not a lot of standardization for building structure creation plans.
    • The roundtable provided valuable input in shaping the selection criteria which worked well for MPAC.
    • Scott Milne also discussed MPAC AI-use cases including property change detection, commercial parking levy, and more.
       
  • AI Communication Plan: Julia Lipman provided a high-level plan for sharing the work of the round table publicly, starting in the fall. A sample draft of the first social-media post was shared. Ways to amplify the message were considered, including proposing a link to a form to collect information from people who may be interested in getting involved. Other ways to expand were also queried including live streaming meetings and sharing documentation.
    • Posting in both official languages will be considered to amplify the message.
• Wrap-up and next steps: 
  • The ALL IN conference is taking place on September 24-25.
  • The Next meeting is scheduled for November 14^th which will align with the G7 GovAI Challenge taking place.

Meeting 4

Attendees: Nicole McNeill, Sam Ip, Simon Hodgett, Scott Milne, Tammy Wells-Garrett, Stratos Kaloutas, Julia Lipman, Soussanna Karas, Mia Cho, Chad Cogar, Mark Daley, Dawn Hall, and Jesslyn Dymond (virtually).  

Meeting Notes Summary: 

• Update on AI Initiatives at MPAC: Tammy Wells-Garrett and Stratos Kaloutas provided an update on MPAC's AI initiatives in 2025, including the development and deployment of the Orchestrator platform, several AI chatbots, the LIR Tool and the Building Plan NLP. 
  • Orchestrator Platform Development: Orchestrator is the underlying system that enables all of MPAC's other generative AI tools. Built fully in-house, Orchestrator features custom-built data pipelines for processing unstructured documentation, endpoints for rapid agent creation, workflow automation, and APIs for integration with other internal applications. 
     
  • Chatbot Deployments: MPAC launched several chatbots, including an IT Chatbot, an HR Chatbot, and a Valuation Chatbot.   
    • IT Chatbot: The IT Chatbot achieved an 83.5% autonomous resolution rate across 4,500 chats, providing 24/7 support and freeing up service desk agents for more complex tasks. Additionally, integration with existing systems allows tracking of recurring issues and provides data-driven insights for continuous IT service improvement. 
    • HR Chatbot: The HR Chatbot is expected to create a 30% reduction in HR inbox enquiries, offer a standardized employee experience, and offer scalability as the corporation evolves. To ensure accuracy, there is a two-tiered system where certain questions can be automatically answered, while more complex questions will have a ticketing system that undergoes human review. 
    • Valuation Chatbot: The Valuation Chatbot is expected to save 15-30 minutes per day, per valuation employee as it enables a more streamlined assessment process and enhances data accessibility. 
       
  • Other AI Deployments: The LIR Tool and Building Plan NLP are two additional AI tools that have been deployed in addition to the chatbots. 
    • LIR Tool: The LIR (Legislation Interpretation and Request) Tool integrates with the legal team's workflow, allowing users to filter requests, access supporting documents, and generate draft responses using AI. It includes features such as semantic search for similar responses and a rich text editor for manual adjustments. The LIR Tool may be scaled up in the future to include additional workstreams.  
    • Building Plan NLP: MPAC developed a custom data extraction tool to process building plans – automatically extracting dimensions and labelling each room. This tool significantly reduces the manual effort typically required to review building plans and enables new use cases for property assessment. 
      ​​​​​​​
  • Governance, Testing, and Quality Assurance in AI Solutions: The roundtable discussed governance, quality assurance, and ongoing monitoring for AI solutions, with MPAC detailing their multi-layered testing approach, feedback mechanisms, and pursuit of ISO 42001 certification. 
    • Multi-Layered Testing Approach: The roundtable also discussed MPAC’s structured testing process for AI solutions, which starts with internal team testing, followed by broader testing with employees who participated in design workshops, and then a final round of holistic testing before deployment. This approach ensures that discrepancies and hallucinations are identified and addressed prior to release. 
      ​​​​​​​
• Update on the G7 GovAI Grand Challenge: Dawn Hall presented on the G7 GovAI Grand Challenge, which focuses on developing innovative and scalable solutions to barriers preventing AI adoption in the public sector. 
  • Structure and Participation: The Challenge will begin with a two-week intensive competition on November 17, 2025 featuring four specific problems for participants to address. Entries will be pre-screened in December 2025, and with judging and solution showcases taking place in early 2026. A youth category is also included to encourage participants of all age groups to submit their solutions.  
  • Supporting Initiatives and Collaboration: Additional initiatives include the formation of the G7 AI network, development of a roadmap for scaling AI projects, creation of a catalogue of open-source and sharable AI solutions that G7 member countries can use, and a collaboration on approaches to measuring AI value and impact. These efforts aim to reduce duplication, share best practices, and facilitate adoption across jurisdictions. 
    ​​​​​​​
• AI Insights: Mark Daley discussed the broader ethical, economic, and societal implications of AI, including resource allocation, the role of government, and the evolving definition of intelligence and human value. 
  • Commoditized Intelligence: The conversation explored how AI is reshaping the value of traditional intelligence, the importance of human interaction, and the need for educational institutions to adapt by emphasizing skills that AI cannot easily replicate, such as collaboration and ethical reasoning. 
  • Geoeconomic Positioning: Mark outlined Canada's strategic advantages in AI, including abundant energy, cold climate for data centers, and a strong talent pool, suggesting that these factors position Canada to play a significant role in the global AI landscape. 
  • Resource Allocation and Ethics: The discussion addressed the ethical considerations of allocating computational resources, such as data center power, between commercial, research, and societal needs. Participants noted the influence of economic incentives and the necessity for government intervention to ensure equitable access and alignment with public interests. 
  • Indigenous Perspectives and Inclusion: Jesslyn Dymond highlighted TELUS's partnership with Indigenomics to integrate Indigenous knowledge systems into AI development. It is important to intentionally create space for and broaden the knowledge systems used both within existing AI models and when building new AI models. 
    ​​​​​​​
• Wrap-up and next steps:  
  • Meetings in 2026 will continue to be held quarterly at Osler, Hoskin & Harcourt LLP on Friday mornings. 2026 dates have been sent to calendars. 
  • Future topics to discuss include identifying the ongoing challenges and potential solutions for the maintenance of AI tools once they are deployed across an organization.  
  • The group will work to Identify opportunities to showcase the MPAC AI solutions with other public sector entities.