Montreal - part 1: Exploring AI needs-centered approach with and for vulnerable communities
Inspired by UN-Habitat’s white paper, AI & Cities: Risks, Applications and Governance, the Urban AI initiative seeks to develop AI projects that are truly adapted to the social and economic realities of local communities in Montréal. Urban AI, known as IA Urbaine in French, is led by the Université de Montréal in partnership with the Canadian non-profit organization Open North. Its goal is to investigate if and how artificial intelligence (AI) can benefit vulnerable and marginalized communities in Montréal.
In this web story, we share the methodology and the lessons learnt from the pilot project in Montréal’s Côte-des-Neiges neighbourhood during the discovery stage, providing an overview of the needs-assessment approach and methodology, the results of the assessment and links with potential fit-for-purpose urban AI systems to be developed and implemented.
The project demonstrates how the AI risk framework presented in the UN-Habitat’s white paper AI & Cities (2022) can be applied in practice, and provides a valuable example of a people-centered approach to AI and community engagement.
A mentoring session to present the pilot project was held on April 24 2024. More information and the recording link are available here.
A second web story is published exploring how AI systems can help address local needs.
Values and principles for responsible and ethical AI in cities
Montréal’s AI ecosystem is thriving. Quebec’s metropolis boasts cutting-edge infrastructure, being the largest AI research hub in Canada, and attracts many major international companies and world-renowned researchers. The city hosts more than 22 000 university students in AI and more than 30 000 professionals with skills related to AI. However, as in other major cities adopting AI around the world, there are still few initiatives that leverage community input in AI to address local needs.
The Urban AI – IA Urbaine project is committed to the responsible and ethical design and implementation of urban AI solutions. There are three key values and principles that guide our approach: 1) rejecting techno-solutionism; 2) promoting equity, diversity and inclusion; 3) respecting the Montréal Declaration on Responsible AI. Let’s dig a bit deeper into these concepts:
- Rejecting techno-solutionism: we recognize that technology, including AI systems, are not always an effective nor an appropriate solution to complex, wicked problems. This means that the project prioritizes residents’ concrete needs and tries to avoid taking for granted AI’s effectiveness in addressing those needs. In essence, we begin by identifying real neighbourhood needs and assess whether AI systems are an appropriate solution, as opposed to beginning with an AI solution and trying to find a corresponding neighbourhood need.
- Equity, Diversity and Inclusion: AI in cities must be fundamentally multi-disciplinary. This is why Urban AI seeks to include the perspective of a variety of stakeholders in order to understand the problem space and identify equitable solutions. The project aims to establish a robust, enduring dialogue among community, technological, political, industrial, and academic stakeholders and develop practical partnerships with local communities in Montréal.
- Respecting the Montréal Declaration on Responsible AI : the project is committed to the implementation of the Montréal Declaration on Responsible AI, a document co-created with citizens and experts in 2018, that features 10 principles designed to guide ethical, inclusive, and beneficial AI use.
A first pilot project in Montréal’s Côte-des-Neiges neighbourhood
Urban AI kicked off in the spring of 2023 with a first pilot project in the Côte-des-Neiges neighbourhood located in the center of Montréal. This neighbourhood was chosen primarily because it is home to the main campus of Université de Montréal, which grants it privileged status due to long-standing ties with the community.
Côte-des-Neiges is a truly multicultural neighbourhood, where people from different ethnic and social backgrounds come together and contribute to the social fabric and cultural vitality of the community. Indeed, more than half of the residents are members of visible minorities, and almost half of them have a mother tongue other than French or English. In fact, a large part of the neighbourhood’s 100,000 residents are immigrants. Like many immigrants around the world, some of them face specific challenges in terms of economic and social integration.
The neighbourhood is going through a period of pronounced change, with some areas becoming increasingly gentrified and others continuing to face challenges in terms of access to good quality and affordable housing. At the same time, neighbourhood residents’ incomes are, on average, lower than in the rest of Montréal. All these factors bring their own set of challenges in terms of the ability of institutions and community organisations to meet the social and economic needs of the population.
Discovery stage - a mixed, participative research methodology
The pilot project began with a discovery stage to better understand the specific and priority needs of local residents, with a focus on more vulnerable or marginalized communities. In this initial phase, Open North, a Canadian organization specializing in data governance and digital strategy, conducted a neighbourhood needs assessment to better understand local needs and the context in which they emerge.
In collaboration with public and community stakeholders, Open North meticulously cross-referenced quantitative and qualitative data in order to create a comprehensive and balanced neighbourhood profile consisting of 1) an accurate, up-to-date socio-economic profile of neighborhood districts, and 2) the identification of several priority needs of the neighbourhood’s vulnerable populations, from an equity diversity and inclusion perspective.
Although the scope of this pilot project did not allow for large-scale public consultations, Open North was able to leverage past and current consultation initiatives through a literature review. In addition, Open North conducted 12 interviews with representatives from various key neighborhood organizations specializing in fields such as housing, employment, transportation, culture, and others. Throughout this process, the project team adhered to the ethical principles of informed consent, confidentiality, and anonymity. After each interview, the interviewee was given a summary of the discussions so that he or she could validate his or her statements.
A qualitative analysis of these interviews was then cross-referenced with information from the literature review as well as quantitative data analyses (primarily from the Canadian census of the 2021 population) to define the primary themes that would serve as the foundation for the needs assessment:
1) better access to information and increased participation in community life,
2) improved access to food resources,
3) finding suitable and affordable housing, and
4) enhancing mobility.
Steering committee – an inclusive mechanism and safeguard to assess community needs
A steering committee was established early in the data collection phase. The committee was composed of three representatives from prominent community organizations and a person responsible for community partnerships at the Université de Montréal. It shared relevant resources, such as existing reports or practical feedback from the field, and identified blind spots in the analyses conducted. As the study neared its conclusion, committee members contributed to prioritizing the identified needs through a collaborative workshop led by Open North.
One of the key learnings in the discovery stage relates to the steering committee, which proved to be an effective monitoring system, especially because it was established early in the project. It played a central role throughout the study, offering valuable advice and guidance while ensuring the inclusion of diverse stakeholder perspectives. Additionally, it facilitated access to resources and key neighbourhood stakeholders. Similarly, collaborating with an intermediary who could bridge the gap between researchers and the field, in this case the Université de Montréal’s team responsible for community relations, proved to be essential. This enabled the precise targeting of relevant contacts and the establishment of trust-based relationships with them.
Overview of the results from the needs-assessment study
The study highlighted the linguistic and cultural diversity of the area, revealing the magnitude of challenges faced by public institutions and community organisations in meeting the social and economic needs of people, especially in a context where a large portion of residents are immigrants. Existing language barriers exacerbate communication and information challenges in this context. Moreover, a double transition seems to be underway in the neighbourhood, with areas tending to gentrify on the one hand, and others where household incomes remain modest on the other. Following the consultations, the 4 primary themes that were suggested at the start of the study were confirmed and explored further. They became 9 specific needs, prioritised as follows:
- Increasing the supply of affordable and adequate housing;
- Reducing language barriers;
- Reducing digital divides;
- Informing tenants of their rights and obligations;
- Improving the discoverability of local services;
- Reducing the reliance of low-income households on food aid;
- Improving the frequency of public transport through data collection;
- Making pedestrian routes, cycle paths and curbs safer;
- Reducing the reliance of community organisations fighting food insecurity on food banks.
A brief overview of links to potential AI systems
Following the needs-assessment study, Open North produced an action plan to guide the development and deployment of AI systems that are responsible, inclusive and fit-for-purpose to the needs prioritised by the neighbourhood’s population. The action plan began by analyzing the links between the 9 specific needs identified and different AI applications, including risks associated to each concrete example.
Based on the UN-Habitat’s framework, the AI applications are listed according to key sectors for intervention in cities: energy, mobility, public safety, water and waste management, healthcare, urban planning and city governance.
The analysis revealed, in the context of this study, three main scenarios:
- AI has limited potential to address certain needs (for example, to reduce digital divides or increase affordable housing, where non-AI or even non-technological solutions are deemed more effective).
- In some cases, effective AI solutions exist to meet an existing need, but the risk is considered too high for the responsible development of AI (for example, AI systems to sense the home environment as a means to detect humidity levels or maintenance needs, leading to significant concerns over surveillance and privacy rights).
- In other cases, AI systems were seen as offering promising applications to holistically tackle local needs. For example, Large Language Models (LLMs) can help reduce language barriers, improve the discoverability of services, and inform tenants of their rights.
Limitations and mitigation best practices – addressing risks to a people-centred approach
It has been crucial for the pilot project to identify and acknowledge certain limitations – see table 1. Despite these limitations and biases, the neighbourhood needs identified were validated through various feedback loops from multidisciplinary stakeholders, giving us confidence that the results, while far from being exhaustive, are anchored in the neighbourhood’s reality.
Table 1
Discovery stage | Limitation analysis | Mitigation best practices |
Selecting participants | Interview participants introduced bias by focusing on four specific themes, potentially excluding other relevant topics | The advisory committee was engaged, and additional documentary and statistical data from external sources were integrated to enhance the robustness of the analysis |
Collecting information | Using community organizations as intermediaries carries the risk of distorting information due to their subjectivity and agendas, which could influence data collection and interpretation | The statements gathered were cross-referenced with the University’s team responsible for community relations, as well as with some of the district’s elected representatives during a meeting at which the results of the neighbourhood study were presented |
Data interpretation | Data interpretation, whether quantitative or qualitative, may be influenced by the subjectivity of the researchers, introducing a potential bias into the analysis | Interviewees were actively involved in validating information, and the advisory committee and project partners were consulted to diversify perspectives and minimize subjective biases among analysts |
Final reflections
This discovery stage does not attempt to be exhaustive or to cover all aspects of the neighbourhood. That would require consultation and research processes on an entirely different scale. Nevertheless, the results obtained through a rigorous research method and the involvement of a wide range of local stakeholders have provided a solid basis for the continuation of the project. The variety and the commitment of the stakeholders was particularly decisive for the success of this first stage. The next steps for the project will involve developing responsible and fit-for-purpose AI tools for and with local residents. This will be the focus of an upcoming story.
![IG on PCSC Flyer (2)[48][15]](https://digitalhelpdeskforcities.org/wp-content/uploads/2024/10/IG-on-PCSC-Flyer-24815-1-e1729604743668.jpg)
