Inclusion
Smart cities are for everyone. Municipalities embracing smart city tools and solutions may focus on big projects (like road and highway infrastructure) and enabling citizens who already own technological tools (such as smart phones and home computers). The problem with this approach is that it does not consider who might be left out of benefiting from these improved services. Not everyone owns a car or a smart phone. Similarly, smart technologies often presume that all users are similarly abled, when we know that this is not so. Persons with disabilities face multiple layers of marginalization and require specific services to participate in civic decision-making.
We should always ask of every smart city proposal and initiative: who is left out? Just as a municipality must serve all who live within its boundaries, a smart city must similarly benefit all. Benefitting all doesn’t just mean pulling everyone under the same umbrella – in some cases, there will be groups who require additional tools or investments in order to facilitate equitable participation. This is not simply a question of who is left out of the smart city: rather, inclusion principles ask whether a technology or approach reinforces or replicates existing inequalities, or even creates new groups who are treated inequitably.
Fortunately, there are many smart city solutions and technologies that focus on accessibility and inclusion. Some examples include:
Free municipal WIFI. Municipalities cannot use smart city tools to reach the disability community if residents lack access to the internet. Adding WIFI routers to public areas that are accessible to the disability community will improve online access to services.
Digital accessibility maps. Municipal maps driven by resident-contributed data can rate the accessibility of different areas in the municipality. Residents will have access to information on accessibility such as whether a particular location has a ramp, braille text, and automatic doors.
Service delivery telephone systems. Residents may dial in for information on municipal services such as transportation, request accommodations for public events, and be updated on changes to municipal disability policies.
Customer Service Applications. Designing applications that combine multidisciplinary professional services (legal, medical, or social) can resolve the multi-layered issues people with disability face without having to redirect the resident to a variety of services. The application can screen an issue and identify what type of professional the resident needs. These applications can also assist the wider resident community as anyone can utilize this service.
Algorithmic bias
What is an algorithm?
Artificial intelligence is becoming an increasingly common feature of municipal governance tools and services. “Artificial intelligence” tools use algorithms - computational techniques and processes - to perform tasks that require intelligence, such as pattern recognition, computer vision, and language processing. There are a variety of such tools, but they roughly fall into two categories: knowledge-based systems and machine learning systems.
Knowledge-based systems deduce behaviour on the basis of set rules - the “inference engine” - and identified facts - the “knowledge base”. The most common knowledge-based systems are “expert systems”, tools designed to replicate how human experts make decisions. Common examples include speech translation systems, machine diagnostic systems and code debugging software.
Machine learning systems, in contrast, use statistical analysis to continuously improve decision-making performance through a feedback mechanism. Machine learning systems build a mathematical model based on sample data, known as "training data", in order to make predictions or decisions without being explicitly programmed to do so. The decisions made by such systems are fundamentally complex and conditional and not transparent. For that reason, these decisions can be impossible for designers to “reverse engineer” or explain. Common machine learning applications include computer vision, self-driving vehicles and facial recognition systems.
Machine learning techniques can augment knowledge-based systems with feedback mechanisms to both improve the inference engine and augment the knowledge base.
Algorithmic Bias
Both knowledge-based and machine learning systems rely on data and can easily perpetuate existing patterns of bias in that data. Further, that bias is obscured by the presumptive objectivity of the algorithm. However, any system designed by humans will reflect human choices and the assumptions, biases, and blind-spots we all hold. This means we need to appreciate that artificial intelligence systems are not neutral, and at the most extreme can produce results that are inadvertently or even intentionally biased, discriminatory, and in violation of human rights.
Fortunately, potential problems inherent in artificial intelligence systems have been recognized. For example, the 2018 Toronto Declaration Protecting the Right to Equality and Non-Discrimination in Machine Learning Systems calls on governments to develop effective remedial mechanisms for those adversely affected by these systems.
How to combat algorithmic bias
Municipalities looking to procure artificial intelligence systems need to be conscious of the potential for biased or discriminatory outcomes. There are a number of measures that may be taken to address these potential problems.
Algorithmic Impact Assessments. In 2019, the Canadian Government adopted a Directive on Automated Decision-Making that requires decision makers to conduct an algorithmic impact assessment to ensure that bias issues are identified and addressed in the production of automated decision making processes that use artificial intelligence to make external decisions that impact clients. The Canadian government has prepared a beta version of an Algorithmic Impact Assessment that can be of assistance in undertaking this initiative. While proactive assessment of human rights impacts is best practice, operational analysis of outcomes of the use of artificial intelligence systems is important, as well. For machine learning systems, such analysis is crucial since the “black box” nature of these systems often obscure how the systems actually make decisions.
Use of Trusted and Independent Auditors. Municipalities wishing to employ artificial intelligence tools may fear they lack the competence to undertake impact assessments. Third party auditors possessing the necessary technical familiarity with artificial intelligence systems may be employed to review both proactive and outcome impact assessments. Such auditors may act as trusted intermediaries to audit artificial intelligence systems, especially when they are owned by third party service providers and their algorithms regarded as a trade secret.
Transparency. Making training data or even an expert system’s knowledge-base publicly available can permit third parties to inspect the data for potential bias, discrimination or equality issues. Transparency initiatives must be consistent with both individual privacy rights and third party intellectual property rights in the data. Follow good privacy practices and review intellectual property issues.
Digital Divide
Municipalities considering smart city solutions should understand how those solutions may affect Canada’s digital divide. “Digital divide” has traditionally referred to the divide between those who have access to broadband connectivity and those who do not. However, over time the digital divide has come to signify the broader discrepancies between those who have access to digital technologies and are able to benefit from them and those who don’t. Aspects of this broader conception of the digital divide include internet connectivity, access to internet-based services, access to devices and access to benefits of information technology applications.
The root causes of digital divide issues are complex, and range from income disparity to societal inequality to lack of digital literacy or even familiarity. Geographical features can also play a role, with urban and more populous regions enjoying greater access to the internet and digital services than rural and sparsely populated regions.
Municipal smart city applications will certainly have implications for digital divide issues. We address inclusion issues at length elsewhere in this website.
How Can Municipalities Exacerbate Digital Divide Issues?
Smart City solutions have the potential to entrench existing inequalities in Canada, including those relating to the digital divide. For example, smart city solutions that focus on cars will tend to benefit drivers over non-drivers, benefiting those with the economic resources to own and drive cars over those who do not. This is not to say that municipalities should shun such solutions. Rather, municipalities should approach smart city planning with a view to benefitting all of its members. For example, traffic planning solutions that benefit car owners and drivers should be part of a wider transportation plan that meets the needs of all. Public transit, ride-sharing and bicycle planning activities all enjoy smart city applications that go some way towards spreading the benefit of smart city approaches across the digital divide.
Similarly, Internet connectivity plans that provide the Gigabit speeds of fibre to a community will only benefit those who can afford a connection unless proactive steps are taken to provide public and accessible connection terminals. Again, community fibre is a good thing; adopting approaches that ensure its benefits spread across community and perhaps “fill in” or at least mitigate the digital divide makes that good thing better.
Smart City Solutions Can Help Overcome Digital Divide Issues
Smart City solutions can also positively address digital divide issues. Inclusion is a fundamental issue for all Canadian communities, and transparency, outreach, consultation and collaboration among municipal decision-makers and the populations served will help ensure that smart services will cross the divide to meet the needs of all communities.
Internet connectivity issues are at the core of all digital divide issues: without access, those on the wrong side of the digital divide cannot hope to enjoy the benefits of digital services. Accordingly, smart city approaches that improve internet connectivity for the young, the rural, those less well-off and the homeless can help overcome digital divide issues. Smart internet connectivity approaches include municipal Wi-Fi systems and public access programs through, for example, the public library. More investment-heavy approaches involve investment in physical infrastructure.
Considering digital divide issues more broadly, municipalities can take steps to ensure that their smart city services are accessible across the digital divide. Digital literacy initiatives can help ensure that all may benefit from smart services. These kinds of approaches are often described as “Smart Citizen” initiatives and characterize the “Intelligent Communities” movement. Similarly, municipalities may approach smart city planning with express consideration for rural and low-income populations, ensuring not just accessibility of services for these communities, but with a view to identifying and meeting their unique needs.
Finally, municipalities should consider what steps they might take to ensure that the benefits of smart city initiatives bridge the digital divide. Investment in the internal infrastructure of municipal operations offers one example: all benefit from improved efficiency of internal operations.
Resources
Guides and Toolkits
G3ict, “Smart Cities for All Toolkit”.
A set of tools intended to help government officials, policy makers and others design smart cities with a focus on accessibility of information and communications technologies by persons with disabilities and older persons, including, for example, a checklist to follow when developing a public procurement policy for technology.
Citizen Lab, The guide to inclusion in e-democracy
In this guide, the Citizen Lab offers principled but practical approaches to ensuring inclusivity designing digital participation platforms. The dcoument offers guidelines for how to communicate inclusively, build an inclusive questionnaire, combine offline and online participation, and ensure representativity.
Treasury Board Secretariat, Directive on Automated Decision Making
The Government of Canada has adopted a directive on how to utilize artificial intelligence to make, or assist in making, administrative decisions in a manner that is compatible with core administrative law principles such as transparency, accountability, legality, and procedural fairness.
Government of Canada, Algorithmic Impact Assessment
This tool offers a questionnaire designed to help assess and mitigate the risks associated with deploying an automated decision system.
Other Resources
Deloitte, “Inclusive smart cities: Delivering digital solutions for all”
Provides a framework to incorporate and prioritize inclusivity at each phase of a smart city initiative and includes examples of approaches cities have taken to address inclusivity.
OpenNorth, “Open Smart Cities Guide”.
This guide provides a definition of an Open Smart City and discusses community-centric development of smart cities including examples of various smart city initiatives.
Frontiers, “The Concept of Sustainability in Smart City Definitions”.
This paper reviews the definitions of smart cities found in literature and suggests that while sustainability-oriented definitions are prevalent, the current implementation of smart cities tend to prioritize technology before the social aspect.
Aimi Hamraie, “A Smart City Is an Accessible City”, The Atlantic (November 6, 2018).
An overview of digital-accessibility maps and related technologies which attempt to address inclusivity of persons with disabilities, including examples and potential issues.
Global Commission on Internet Governance, “One Internet”.
A comprehensive look at building an open and accessible internet for everyone, including the issue of the “digital divide” between those with and without access to the Internet.
NPR, Automating Inequality, (2018).
An interview with the author of Automating Inequality, Virginia Eubanks, highlighting the (ineffective) use of automated systems in public services.
Cititzenlab, What’s the difference between diversity, equity, and inclusion? (2021).
An explanation of how diversity, equity, and inclusion are three related concepts but are far from synonymous.
Resources - Algorithmic Bias
Treasury Board, Canada, “Directive on Automated Decision-Making”, (2019).
Government of Canada, Beta-version, Algorithmic Impact Assessment.
The Berkman Klein Center for Internet & Society at Harvard University, "Artificial Intelligence & Human Rights: Opportunities & Risks", (2018).
James Manyika , Jake Silberg and Brittany Presten, “What Do We Do About the Biases in AI?”, Harvard Business Review (2019).
Will Knight, “AI Is Biased. Here's How Scientists Are Trying to Fix It”, Wired (2019).
Haro Ken, “This is how AI bias really happens—and why it’s so hard to fix”, MIT Technology Review (2019).
Andreas Tsamados, Nikita Aggarwal, Josh Cowls, Jessica Morley, Huw Roberts, Taddeo Mariarosaria and Luciano Floridi, The Ethics of Algorithms: Key Problems and Solutions, (July 28, 2020).
Resources - Digital Divide
George Sciadas, “The Digital Divide in Canada”, Statistics Canada (2016).
Allison Bramwell, Ken Coates, and Neil Bradford, “Expanding digital opportunity in Canada?”
Nicole Goodman and Zac Spicer, “Winning the smart cities challenge with equity, inclusion”.
OECD, “Bridging the Digital Divide”.
Open Media, “11 things Canada should do now to close the Digital Divide for good”, (2020).
Public Interest Advocacy Centre, “#WCRD2017: Building an Affordable Digital World For Everyone”, (2017).
“The Digital Divide, ICT, and Broadband Internet”, Internet World Stats.
Stanford University, “The Digital Divide”.
Last updated