This May, ECNL held a session at the Computers, Privacy and Data Protection Conference (CPDP) in Brussels. We organised this session in the context of our Framework for Meaningful Engagement that we developed based on input of 130+ colleagues from civil society, academics and private sector. The framework serves as a concrete guidance on how to conduct meaningful engagement in AI impact assessments.
We took this opportunity to have a multistakeholder conversation about the meaningful engagement of people - especially those (potentially) affected - in AI development and impact assessment. During the session we explored the options and opportunities for companies and policymakers to meaningfully engage external stakeholders in tech policy and development.
Our discussion was centered around the challenge of participation in practice: from limited decision-making power of civil society to short product development timelines. Speakers shared their expertise and practical experiences related to stakeholder engagement when developing, deploying, and regulating AI, without falling into the trap of tokenism.
Our speakers highlighted that in order to engage people who are affected - or the proxies of those affected (e.g. CSOs) - meaningfully, one must involve them:
- early in the process,
- take the time to conduct the engagement (instead of one-off meeting),
- consider compensation for their time, and
- recognise their expertise of the context.
In addition, speakers contextualised multistakeholder engagement within the European tech policy ecosystem: what’s working or could work, and what isn't? Adequate stakeholder mapping is still a challenge for policymakers and tech companies. Speakers however reaffirmed its necessity. It ensures that no relevant information is missed. It is also crucial for human rights impact assessments to serve as effective tools in developing human-centered and human-serving technology.
From data protection standards to the EU AI Act and a future Council of Europe Convention on AI, we discussed ways in which we can work towards a truly inclusive digital society, especially for marginalised and vulnerable groups. Promising in this regard, as was noted by one of our panelists:
"Something important in AI Act is that it shows that you can actually say “no” to certain uses of AI. There are so many obstacles to accountability and the risk is so high that some practices should actually be prohibited".
(The discussion was captured by live tweeting. Read the thread for the incredibly interesting insights from speakers.)
We thank our amazing speakers for sharing their knowledge and experience. We will continue working together towards the inclusion of all, especially those most at risk, in the development and deployment of AI. Find the list of speakers below:
- Alyna Smith, Advocacy Officer at PICUM (Platform for International Cooperation on Undocumented Migrants) leading their work on access to justice and access to health for undocumented people, as well as legal strategies.
- Laura Galindo - Romero, AI Policy and Governance Program Manager at Meta, leading programs aimed at improving the quality of rule-making processes in the field of technology policy.
- Jana Gajdošová, Head of Justice and Security Sector in the EU Agency for Fundamental Rights. Jana is a human rights lawyer by background and with areas of specialisation including: digitalisation in the area of justice and law enforcement, rule of law and access to justice.
- Mirko Tobias Schäfer, Associate Professor at Utrecht University's research area Governing the Digital Society and the Department for Information & Computing Sciences. Mirko is the co-founder and Faculty of Science Lead of the Utrecht Data School.