Municipal policymakers need to make educated decisions to shape the future of mobility in cities; there is a persistent assumption that these decisions can be facilitated by a data-driven platform.
The use of disruptive technologies (e.g., artificial intelligence, machine learning, simulation and prediction models, mapping, data analysis) in policy-making raises significant questions when applied in democratic societies that value human rights – and especially when applied by public institutions. These questions involve decision-making (How is AI and other disruptive technologies [e.g., machine learning, simulations] utilised in a decision-making process?); human oversight (How is human oversight ensured?); accountability (Who is responsible for technological outputs, and how can processes involving disruptive technologies be made open?); access & exclusion (Who has access to use disruptive technologies, and what data do large models include?), technology and development processes (Who is included in development and how are models trained?); and trust (How can AI and disruptive technologies be incorporated into democratic policy in a way that is worthy of public trust?).
Urbanite partners organised participatory Social Policy Labs (SoPoLabs) in four pilot cities (Amsterdam, Bilbao, Helsinki, and Messina) to gain insight into the application of disruptive technologies to develop ecosystems of shared mobility data (mobility data commons). Sessions included local policymakers, civil servants, and other stakeholders from civil society. This revealed several challenges and opportunities related to the (lack of) trust of the public servants and end-users in the use of data and disruptive technologies.
The challenges in these steps and requirements provide the context for subsequent recommendations around the leading question of this deliverable: Which factors influence the adoption and integration of disruptive technologies (e.g., artificial intelligence, machine learning, simulation and prediction models, mapping, data analysis) in decision-making processes? Urbanite approaches this question through the lens of municipal efforts to develop local mobility data ecosystems that integrate various datasets and disruptive technologies. Our findings are relevant for all decision-makers (e.g., policymakers and civil servants) working with data and disruptive technology in a public role.
The development of a local (mobility) data ecosystem is a complex process, in which challenges appear at each crucial step. These challenging yet crucial steps are:
- Identify the need for data and use / development of disruptive technologies – It is crucial that real needs of policymakers drive any technical development. The needs and problem that a technical solution intends to solve must be clearly defined at the outset. Some civil servants may be apprehensive to use new technologies because of the time investment, because they are already familiar with existing technology and processes, because of a lack of expertise in developing and implementing disruptive technologies, and because of a lack of trust that disruptive technologies and new ways of working with data will be helpful.
- Define the mission and vision – Based on the needs and desires of the relevant policymakers, establish the mission of the initiative, and review the partners’ and public administrations’ policy and strategy around data use and management.
- Gain awareness of existing data – Once the mission and vision established, development teams need to identify existing relevant data.
Awareness of existing data is a challenge within municipalities for a variety of reasons (e.g., data silos, or a lack of interdepartmental communication) which are often organisational and bureaucratic in nature, rather than technical. Lack of awareness can lead to duplicated efforts and more data collection than is necessary.
Communication between municipal departments is essential to know which data is available in each department and investigate the available opportunities for sharing that data for other purposes that may arise with the integration of new technologies. Municipality should facilitate interdepartmental data exchange, and consider setting up a data management position/team/department within the public administration to facilitate knowledge and data exchange between various (internal) departments and (external) stakeholders.
- Access existing data – Once a team is aware of existing data, that data is often difficult to access.
There are many potential challenges to accessing existing data. Relevant proprietary data may be closed, costly, or gathered by private parties in a manner inconsistent with public values (e.g., gathered via dark patterns or surveillance capitalism). Public data sets may also be difficult or impossible to access, for example due to interdepartmental access restrictions.
Challenges in accessing data are not always technical. Civil servants may be hesitant to use data due to the potential consequences of wrongfully sharing data (e.g., violating GDPR). Apprehensive civil servants may choose not to share the data at all, even in cases that are legally and ethically sound. Similarly, open data sets may not be accessible in practice if civil servants do not know when or how to handle them.
- Ensure quality, cleanness, completeness, and accuracy of data – Once data is accessed, there are various potential reasons why it may be unusable: data may be of a poor quality, unorganised, incomplete, or inaccurate.
Development teams may face difficulty in identifying technical problems which (later) arise due to missing or otherwise problematic data. Problems with data at this stage in the process can lead to subsequent problems in development (e.g., when designing for interoperability and usability).
- Meet interoperability standards – Various high quality data sources must also be interoperable in order to contribute to a larger data ecosystem.
The lack of widely and consistently used open standards for various datasets (e.g., even within municipalities and municipal departments) is a major challenge for city-level data ecosystems and data commons. Data sets are often organised to suit an isolated use, domain, or context, and not optimised for collaborative use alongside other data sets.
Despite these challenges, certain open standards are emerging which have been used in Urbanite.
- Establish a governance framework –After establishing the mission and vision of the initiative and the data needs, requirements and policies, it is crucial to define the governance structure, the governing body, and the roles and responsibilities that will facilitate its implementation. To do so, identify applicable regulations and constraints to data sharing, its exploitation and underlying methods. Keep in mind the rights of the individuals and organisations who generate and consume the data during its whole lifecycle, and apply specific guidelines, methods and tools to ensure its compliance.
Challenges in each of the above steps must be met before technical development of a larger data ecosystem can proceed. Further requirements also pose crucial challenges during development and use of data and disruptive technologies. These challenging requirements include:
- Human decision-makers must understand and scrutinise the implications and limitations of technological outputs (e.g., data visualisations, models, and predictions).
- The use of data and disruptive technologies must be compatible with relevant laws and public values (e.g., regarding privacy and transparency).
- Technological outputs must be relevant for and usable by decision-makers.
- Technological outputs must be explainable to the extent that the steps of reasoning, sources, and considerations that lead to a recommendation or prediction must be able to be reproduced and validated by humans.
- Use of the technology must merit and earn trust by citizens, decision-makers, and other stakeholders.
There are many complex challenges which arise from necessary steps in planning, developing, and using data and disruptive technologies in the context of decision-making and the development of data ecosystems. Such challenges can be overcome; a public design process with the right priorities can lead to the development and use of technology which both aids in human decision-making and earns societal trust.
Based on our experience with the Urbanite project, we have created a set of recommendations for local policymakers on how to deal with the aforementioned challenges. These are presented in the form of design process priorities, and are elaborated in more detail in the Urbanite deliverable “D2.6 Impact analysis and recommendations”. These design process priorities include:
- Get a clear understanding of the internal organisation of the public administration;
- Participatory development with decision-makers, citizens, and the people who are expected to implement the technology;
- Identify a shared mission with the relevant stakeholders;
- Develop the solution in a modular and iterative manner;
- Work open when it comes to standards, technology, and processes;
- Ensure the explainability of the technology and AI;
- Create model trustworthiness by following codes of conduct regarding data ethics; and
- Educate the relevant stakeholders to increase their data literacy.
Data-driven technologies for policymaking can have a large social impact; they are after all disruptive technologies. To ensure a positive impact, it is important to look at two aspects of the social uptake of the technology: informed human decision-making and trust.
Informed Human Decision-making
The purpose of utilising data and disruptive technologies is to help humans make better, more well-informed decisions. However, the limitations and implications of data and disruptive technologies are vast, complicated, far-reaching, and nuanced. Algorithms make mistakes, data sets are inherently biased, and black-boxed outputs cannot be explained; there are myriad reasons why human discretion should always be prioritised and protected when using technology within a public decision-making process. To this end, socio-technical data literacy regarding societal implications of data and disruptive technologies must be a prerequisite for any decision-maker that utilises them.
First and foremost, humans need to make decisions – not AI or other technology.
Beyond this simple tenant, it is also crucial that decision-makers have nuanced understanding of the complex context in which mobility data sits. Without such knowledge, a specific tool will not provide the correct feedback to steer decision making. For example, a simulation to improve traffic flow may recommend to build more roads. Building more roads, however, is contrary to sustainable development goals. Moreover, the simulation is likely to have been fed more data about cars and roads (which is relatively easy to gather) than about more sustainable modes of transportation like walking and cycling (whose data is more difficult to gather). Decision-makers and developers need to avoiding letting data sources determine design decisions, and instead make their own informed choices about what to optimise in their technology and decision-making processes.
Trust about the use of disruptive technologies is a prerequisite for their application in matters of society and governance. Trust applies on many levels in this regard and from the perspective of citizens and decision-makers alike, for example:
- trust that the data is accurate and reflective of peoples’ lived experiences;
- trust that decision-makers are knowledgeable about the nuances and limitations of the technologies they are using, and are comfortable using it;
- trust that decisions are made by humans (not technology);
- trust that privacy and other public values are ensured by the technical systems’ design;
- trust that the technology is open and transparent;
- trust that outputs of AI are explainable valid;
- trust that (human) decisions informed by AI outputs are explainable and valid;
- and more.
It is not possible to achieve trust in these necessary instances without a participatory and educational design process that prioritises socio-technical data literacy and follows the necessary development steps. Public administrations could increase their communication with citizens and stakeholders, through for example organising events and initiatives that include the citizens in the technical project. This helps to eventually close the gap and build trust between civil servants and citizens, which could then increase the trustworthiness of disruptive technology. To quote one Urbanite partner, “There can be no trust without understanding and scrutiny.” In other words, the goal is not to develop trust amongst society (about the use of disruptive technology); but rather to build technology and protocols that merit public trust because of their openness and alignment with public values.
Stakeholders’ concerns and focus are quite similar; they involve trust, public values, human decision-making, openness, and public involvement, and how to address these issues through concrete choices in technical design processes.
There is a persistent belief that data-driven technologies can pose solutions to contemporary urban mobility issues. Our findings point to the need for shared public values to form a foundation that guides urban mobility development through various challenging steps and requirements. Only then can decision-makers ensure that technical solutions are in line with the mission of democratic institutions; and only then can developers have a common ground to align and develop other necessities like access and interoperability.