[Spotlight] Walking the walk of Data Ethics
From theory to practice
By Darine Benkalha*
Despite our increasingly data-driven world, the collection, sharing and use of data still provoke numerous debates. On one hand, public, private and civil society organisations all use data to inform their activities whether it is policy making, research on societal issues or the creation of applications and smart technologies. On the other hand, big data scandals sparked a “data fearing” dystopia where various actors, and especially individuals, practice “data holding”. However, those collecting, sharing and working with data are facing a growing public criticism and more are exploring the ethics of their data practices and adopting ethical frameworks to guide them.
The Open Data Institute (ODI) defines data ethics as a “branch of ethics that evaluates data practices with the potential to adversely impact on people and society-in data collection, sharing and use.” For the ODI, data ethics should not focus solely on personal data but on all types of data and should be used at every stage of the data lifecycle. In fine, data ethics encourages the adoption of guidance that will ensure more ethical decisions are made about data and especially when data activities have the potential to affect people and society.
For our September’s Implementation Working group (IWG), we invited Olivier Thereaux from the Open Data Institute, Sara Baker from The Engine Room and María Paz Canales from Derechos Digitales to discuss data ethics and explore the principles and behaviours we should adopt when we work with data and why all of this matters for the public good.
While in the past years many organizations have taken a first step in identifying codes of conduct for the ethical use of data, in practice, these principles are rarely binding and not systematically embedded in day-to-day activities. Insisting that data ethics shouldn’t be a “check the box” type of exercise, our panelists shared some practical tools and guidance that should help walking the walk of data ethics.
Open Data Institute’s Data Ethics Canvas
Our first panelist, Olivier, introduced the ODI’s Data Ethics Canvas, with the hope to provide a practical framework to anyone who works with data. Indeed, the Canvas main goal is really to help identify and manage ethical issues right at the start of an initiative as well as throughout its whole process. Built around the values of engagement, trust and with the goal of minimizing harms on already marginalized groups, the Data Ethics Canvas counts four focus areas each including a set of concrete asks:
- Know the data (pink) to analyse the existing environment and how the data is acquired.
- Explore impacts (yellow) with a focus on taking into consideration the trade-off between positive and negative impacts both on the short and long term.
- Plan your engagement (green) as communication is key to build trust and help people understand the what, the why and the how of a project.
- Integrate ethical practices (blue) to embed and think continuously about ethical issues and values.
Olivier encouraged future users of the canvas that each of its areas should be addressed by a cross-cutting team of experts from diverse backgrounds, that the canvas should have an equal amount of details and actions related to all focus areas and that it should be regularly reviewed and updated as the project progresses. Olivier also shared tips that practitioners should take into consideration during the data lifecycle:
- Collect: Ask yourself questions about both the legal and the ethics of collecting data.
- Use: Inform people how the data will be used, what the results are and how decisions have been made based on this data.
- Share: data should be available equitably and fairly to those who need it the most. Not sharing data can be as unethical as sharing it.
For more detailed information on ODI’s Canvas, check out Olivier’s slides.
Responsible data in project cycles
The Engine Room,with its cross-sectoral group of 1200 practitioners, has also been advancing the importance of data ethics. Their Responsible Data program brings together a community of people committed to using data more safely and ethically for social change as well as creating space to share challenges and to develop practical tools and resources to address them.
Sara explained to the group that while The Engine Room initially took a rights-based approach to data ethics, it switched to a justice-based one to be cognisant of the wide variety of dynamics at play in the data lifecycle. According to them, responsible data should be rooted in the principles of empowerment — empower users to be active participants and co-design the process with them — and harm avoidance — ensure no harm is done or exacerbated by the way data and technology is used. Like the ODI, The Engine Room has identified a series of questions and factors that should be taken into consideration at all the stages of a data-driven project cycle:
- Think about the power dynamics that are at play between the subject and the user of the data.
- Avoid unknown unknowns by practicing ongoing risks assessment.
- Adopt precautionary principle to ensure harms can be sufficiently evaluated.
- Embrace the spirit of thoughtful innovation instead of acting for the sake of rapidity.
- Remember about diversity and bias to assess what perspectives are missing.
- Work towards building better behaviours as responsible data is not one-size-fits-all and is all about context and culture.
Sara also shared with participants more tips to use at each phase of the project cycle:
- Design: What kind of data do you need?
- Collection: Have you obtained informed consent?
- Transfer & Storage: Where do you keep your data?
- Analyzing & Archiving: Has your data been de-identified and anonymized?
- End of Life: When you close a project, what will happen to the data?
See more in Sara’s Slides here.
Data ethics in the health sector
While health sector organizations have been doing some thinking around the collection and use of data for a long time, the recent use of technology for exposure tracking in the COVID-19 context has accelerated the need to have these conversations. In March 2020, the World Health Organization (WHO) created a special task force within its pre-existing group working on guidance for the ethical use of AI in the health sector, to create principles for ethical COVID-19 data management. María was part of this task force representing her organization, Derechos Digitales, and therefore provided the IWG with two key considerations for the adoption of data ethics principles in the health sector.
- Be mindful of the context: creating guidance should always be done with the normative, social and political dynamics in mind as principles will change depending on the beneficiaries’ realities as well as the type of data infrastructures in place.
- The devil is in the details: the number of principles to adopt, the system’s implementation time limitation and the criteria to assess how principles will be implemented are all key to a successful adoption of data ethics principles.
María also shared some recommendation around the adoption of data ethics principles:
- Use restrictions: data should only be used for the communicated purpose.
- Increase the level of responsibility: ask data users to provide information in a way that is understandable for any users of the system.
- Accuracy: only collect and use data that is relevant to what your system intends to deliver.
- Accountability and oversight: consult and engage external stakeholders to co-create the principles.
Data Ethics: Making the most out of data
Olivier reminded the group at the start that data ethics should not only be about avoiding scandals, but also about creating trust when working with data correctly. Indeed, data and trust go hand in hand and the only way we will get the best value of data is when we provide the data to the people who need it the most. As data keeners, we also must continue promoting the adoption of data ethics principles as this is a central part of strengthening public trust and going against the forces discouraging data collection, sharing and use.
But we also have collectively more work to advance on that front. We must continue to advocate for a better balance between ethical guidelines, international frameworks for human rights as well as local regulations. Bridging the gap between international human rights and new technologies guidance will be fundamental to make data ethics principles relevant and eventually binding.
We also need to take the conversation beyond privacy. While it is still the entry point of data ethics, studying the impacts of the use of data on different groups demonstrates that data ethical questions touches broader topics like human rights, as well as very tangible issues like access to health which is of particular urgency in the pandemic context we are all experiencing today.
To sum up, by walking the walk of data ethics, we help mitigate the negative impact and biases that have been unintentionally built into systems, policies and services–to create fairer societies. This is where the true power of data is.
*Darine Benkalha is an IWG participant who volunteered to write this blog. She acts as a Policy Analyst on the Open Government Team of the Office of the Chief Information Officer from the Treasury Board of Canada Secretariat.
Our Implementation Working Group sessions are held monthly. The recording of this one, can be found here. If you would like us to add you to the mailing list, please send us an e-mail at firstname.lastname@example.org.