• Research Brief

AI Futures Policy Lab: Toronto Pilot Summary

by Sarah Villeneuve
Aug 30 / 18

AI Futures Policy Lab

On June 25, 2018, CIFAR and the Brookfield Institute for Innovation + Entrepreneurship (BII+E) hosted an AI Futures Policy Lab in Toronto.

This lab was designed to facilitate capacity building for emerging policy leaders, both within and outside of the civil service, by encouraging critical thinking surrounding a number of possible future AI scenarios based in 2028. In order to create an intimate space that enabled thoughtful collaboration and information sharing, attendance was capped, with total of 18 participants present on the day. The workshop incorporated foresight exercises with brainstorming activities to develop contemporary AI policy approaches in a variety of domains. This was the first in a series of five workshops that will take place across Canada throughout the remainder of 2018.

Policy Lab Activities

1. The AI Thing From the Future

The workshop kicked off with a card game, The ‘AI’ Thing from the Future. The purpose of this activity was to encourage participants to be creative and set the tone for the proceeding exercises throughout the day, which pushed attendees to think beyond our current reality. Participants split into four groups of five, each accompanied by a facilitator. Each group was then given five cards, each containing a different prompt: ARC, to signify what type of future; terrain, defining the thematic context or location of the object; object, specifying the type of artifact you are focusing on; mood, suggesting how you might feel when experiencing this thing; and AI, indicating the technological capability or application that needs to be integrated in your future “thing”.

Each participant was provided with a template to record their idea. Participants were required to individually imagine a future object, or ‘thing’, utilizing all card prompts. Each participant then had the opportunity to share these ideas with the rest of the group.

2. AI 101

Following this exercise, participants were joined by Katya Kudashkina, a researcher from the Vector Institute, who defined common AI terms, described learning techniques, and provided expert knowledge on AI applications across a variety of fields. Participants were able to ask questions and receive clarity from a professional working in the area of AI. This provided attendees with knowledge and resources to draw on for subsequent activities.

3. Future AI Scenarios: Canvas Exercises

Participants were divided into pre-assigned groups, curated with the purpose of bringing together individuals from different policy domains. Each group was joined by a facilitator who presented 3-4 case studies. Groups were given 10 minutes to select one case study, which would become the focus for the subsequent three exercises. Each case study card was developed to reflect a possible AI scenario in 2028, ranging from prescriptive legal analytics to smart homes.

Once the case studies were selected, groups received the first canvas, prompting them to consider the following issues:

  •   who would be affected in this future AI scenario and how;
  •   the impacts it would have at both the local and global level (using a STEEPV approach);as well as
  •   what policy domains would be affected and how.

Each group was given 45 minutes to collaboratively fill out the canvas. Once the time elapsed, groups gave a quick overview of their case studies and the ensuing deliberations.

After lunch, participant groups were presented with the second canvas, structured as a backcasting exercise. This activity required participants to map specific developments expected to occur between 2018 and 2028 that were needed to make their AI future scenario a reality. This exercise was also structured using the STEEPV framework. Participants were encouraged to openly discuss each category.

Upon completion of the second canvas, each group was given the third and final canvas, as well as a deck of cards. Each card represented a different style of government intervention that could be applied in relation to the future AI scenario. Sample cards shown below:

4. Toolkit Development

Groups were then given the opportunity to reflect on the day’s discussions with the goal of formulating a “toolkit”. The aim of creating this resource was to outline what questions to ask or considerations to take when assessing the potential use and implications of an AI technology in a particular policy domain.


While the workshop was not intended to develop recommendations for actions, there were some common approaches and themes that arose across groups:

  •   While AI developments will require new policy solutions, the policy process wasn’t seen to require drastic change
  •   There is considerable value in convening multistakeholder groups early in the policy process
  •   Where possible, it is worthwhile to consider the entire spectrum of policy levers and begin with lighter touch options
  •   It is important for all affected stakeholders to better understand the specifics of a given AI implementation, rather than its technology in the abstract

General Remarks

Participant feedback indicated that it was beneficial to have the opportunity to collaborate with fellow professionals in the policy space. Additionally, many attendees highlighted the foresight approach as a useful framework for thinking through future policy implications of AI and the type of government responses necessary today. However, participants also expressed a desire for more introductory presentations on AI to help ground discussions, and suggested that it would be valuable to discuss real-world AI use cases, as opposed to hypothetical future scenarios. Additionally, feedback noted that having more attendees from other sectors (e.g. private, not-for-profit, and academic) would contribute positively to the collaborative experience.

Overall, this workshop empowered attendees to ask critical questions regarding AI techniques, applications, and potential policy implications.

Next Steps

In September, CIFAR and BII+E will host the second AI Futures Policy Lab in Edmonton, Alberta. Taking into account the feedback from the Toronto Pilot event, a new agenda has been designed to incorporate more introductory presentations, as well as the analysis of existing AI applications, alongside future scenarios. This lab will also require participants to formulate policy briefs with the aim of generating greater awareness of current AI capabilities and applications, as well as the necessary and applicable government responses.

Read the Full Report