In September 2023, Connected by Data brought together a diverse group of civil society, government and academic stakeholders on the fringes of the 2023 Open Government Partnership Summit in Tallinn, Estonia, to co-design model policy commitments that could deliver meaningful transparency, participation and accountability in data and AI governance.
The workshop developed three commitment drafts, focusing on:
- Participatory oversight of technology procurement through creation of multi-stakeholder oversight groups that are empowered to advise, question and publicly report on public sector data and AI procurement.
- Deliberative development of data and AI strategies at both national and local levels through use of both open, and more targeted, participation processes.
- Strengthening citizen voice within sectoral regulators to create a robust feedback loop and ensure data and AI firms have social licence to operate in the way they do.
This Connected Conversation shared these commitment areas in more detail, and invited responses from a range of experts to explore the opportunities and challenges for putting these ideas into practice.
Invited responses
We started the conversation with three invited responses:
Adam Zable, Director of Emerging Technologies at Digital Trade and Data Governance Hub, shared work on mapping data governance globally, which compared the extent of public participation in developing National AI Strategies across 69 jurisdictions. Adam described how
- Of 43 countries which had a strategy, 18 had some public outreach in the process of its development.
- Of these 18 countries, few went beyond consultation into deeper public engagement, and few governments shared evidence of how public input informed their strategy.
While no country can be held up as a model of responsive democratic decision making around AI, Adam outlined some best practice/good examples that can be seen in ongoing consultative processes, such as Malaysia’s AI roadmap designed as a living document to be continually updated.
Adam also pointed to Chile as an exemplar for participatory process around AI, with their extensive engagement of citizens during the development of the first National Policy on Artificial Intelligence, and engagement positioned on the spectrum of participation towards involvement rather than simple consultation. Chile organised regional roundtables to obtain citizen feedback throughout the country, and the government also let people organise self-convened roundtables: sharing results online on an open platform.
Tonu Basu, outgoing Deputy Director of Thematic Policy Areas at the Open Government Partnership, emphasised how action on participation around AI is an important aspect of the wider OGP digital governance conversation, and work on AI strategy is very timely. The new OGP strategy encourages ambitious roadmap thinking to add up to larger impacts over 5 years, rather than a commitment by commitment approach, and work on AI and data governance should fit this approach.
Tonu outlined how effective commitments on AI will require cross agency convening and mandate, and it is important to include this in the commitment design. A siloed agency cannot convene across a government. Even in specific processes such as procurement, or within sectoral regulators, there need to be links built in across government for their work to be effective.
Tonu raised the debate to be had about algorithmic accountability policies, and whether and when these should be binding or not. Some countries have optional opt-in approaches, but this could also be approached through building on data protection law. Tonu suggested that it has yet to be seen, and will require more thought, about what is the most suitable approach.
It can be easy to say “we will establish a multi stakeholder group”, but this needs more detail. How do we make sure that citizens and affected communities are brought in effectively and thoughtfully, and how do we make sure officials are empowered to do that? At what stage of strategy do we best involve the public? Clear milestones are needed to set this out.
Kristen Robinson, Head of Advocacy at Open Contracting Partnership, focused her response on procurement, as an area where we can have goal-focused and meaningful participation. She noted that the goal focus is essential for effective participation, to avoid participation fatigue.
However, Kristen argued that setting up a single tech procurement advisory committee as proposed in the draft OGP commitment may not be realistic given the very broad application of different technologies across different public services. One stakeholder forum would be unlikely to be able to cover all of procurement in this space.
Where stakeholder engagement could be more effective may be in pre market engagement with suppliers; this would be a good moment for more stakeholders to be brought in. Following this, it would be ideal to build relationships so ongoing implementation – during which new issues may emerge – has stakeholder engagement throughout. More broadly, it may work best to seek processes that already exist within which public participation can be added.
Furthermore, Kristen noted, staff training is essential to keep up to date with the implementation of changing technologies, add-ons etc., and flagging key changes during implementation that should be subject to consultation or public engagement.
Some of the best practices to learn from in procurement have been driven by patient rights organisations, parents groups and so on - people who have a stake in the outcomes. The procurement officer is unlikely to see the best options alone, and this demonstrates the value of participation.
Discussed recommendations
Following our invited responses, we heard wider responses to the draft commitments. Some recommendations included:
- A need for building public capacity to engage, building from the grassroots up to push for proactive change.
- A need for success stories, exemplars, and inspiration.
- Thinking about data as infrastructure to find useful analogies and learn from other processes, such as planning frameworks.
- More exploration and understanding of what prevents public participation, for example governments’ innovation focus, corporate capture, etc.
- Broader learning from other areas, such as environmental law, to support participation advocacy in AI and data governance.
Refining recommendations & supporting action
We were delighted to have the opportunity to gather constructive critical feedback on the outputs of our Tallinn Design Lab, and we’re looking forward to continued work exploring how Open Government movements, and movements for collective, participatory and powerful public governance of data and AI can work together.
If you are working on an Open Government action plan, whether nationally, or as part of OGP local, and you want to talk about ways to develop an effective commitment around the governance of data and AI, we would love to hear from you.