IGF 2022 Town Hall #83 Public interest algorithms for content recommendations

One of the key aims of this IGF edition is to discuss opportunities, challenges and risks linked to the ever-increasing use of advanced technologies in our societies. The IGF rightly spots that multi-stakeholder dialogue and cooperation are required to ensure that these technologies are developed and deployed in a human-centric and human rights-based manner. With our session, we want to contribute to this approach by discussing with a variety of stakeholders likely ways to cooperate in order to build an algorithm for content recommendations optimised to achieve public interest objectives. Currently, algorithms for content recommendations have a huge impact on the way people access and share content, and thus on their information diets. Therefore, this advanced technology has a major impact on media diversity policy objectives, which are a fundamental pillar of a democratic society. Being offered by private providers, and in primis large social media platforms, the existing automated recommendation systems are optimised for goals that focus on profit-making and take into no account the public policy objectives that are linked to them. This scenario, which as known has raised numerous challenges for people and society alike, can and must change. However, no stakeholder alone might be able to make this change happen. For example, large platforms have no incentives to change their automated recommendation systems into ones that produce less monetizable outcomes. Alternative players cannot easily enter the market, which presents high barriers to entry. The focus of this session is to discuss who, among the various stakeholders, could lead on or contribute to the design and development of algorithms for content recommendations optimized to achieve public interest goals rather than strictly profit-oriented ones, such as diversity of content exposure. The discussion will touch upon the legal, technical, economic and societal challenges that such a project could entail. It will also include a reflection on the role that civil society, app developers, large platforms, regulators, and the state should play for public interest algorithms for content recommendations to become a reality in our communication and media environment. The discussion will kick off from a related research project performed by experts within ARTICLE 19, which also includes semi-structured interviews with relevant stakeholders.

We plan to alternate onsite and online panellists’ interventions and panellists/audience interactions throughout the entire discussion. The onsite moderator and online one will share this task and make sure that the time is allocated fairly and efficiently to this purpose. The onsite moderator will focus on managing the panellists’ interaction and the interactions across the audience onsite, and the online moderator will focus on managing the online chats/Q&A and will act as a bridge between the onsite and online communities of participants. As we plan to alternate onsite and online interaction, this structure will be easy to follow as some panellists will attend in person and others will join online. We will use the same structure for the interventions from the audience. The aim is to shape the session as a common space where the online and onsite participants can meet and have a smooth exchange and a fluid conversation. We plan to use the chat and Q&A tools to make online v onsite dialogue easier and smoother.