Project

Publicly funded and governed AI infrastructures – A pathway towards just transitions?

The PublicAI project is working for the Mozilla Foundation on the question of whether and how public AI infrastructures represent an alternative form of governance that can better promote research, innovation and learning with AI. Against the backdrop of the monopolisation of large social platforms (google, tiktok, amazon, etc.), a similar development with AI should at least be reconsidered and mitigated. This is because the privatisation of digital infrastructures leads to less transparency, mutual learning and market entry. The project relies not only on an extensive literature review, but above all on an explorative qualitative research process based on semi-structured interviews with a large number of different relevant actors in Europe and the USA. The aim is to discuss the extent to which forms of public AI already exist, what can be learnt from them and what further public AI infrastructures could potentially look like.

Experience with social platforms (google, tiktok, amazon, etc.) has shown how the private organisation of digital infrastructures can lead to great market power and the exclusion of outsiders from relevant knowledge and data. It was not until many years later that questions of data availability, the interoperability of platforms and the opening up of algorithms for mutual learning and research were discussed. In the case of artificial intelligence (AI), similar framework conditions exist: Network, scale and learning-by-doing effects mean that large AI structures in particular are especially competitive, which in turn also means market power and possibly again the tendency to exclude potential competitors and researchers. Therefore, the project aims to explore alternative governance models of AI and how they can better incentivise research and innovation with AI. Public AI infrastructures can exist in different ways, from transparent governance of a privately organised AI to the public provision of an AI and the necessary infrastructure.

The project is based on a review of existing literature on public AI infrastructures, primarily in the EU and the USA. In addition, experiences, challenges and contributions of public AI infrastructures for research and innovation will be collected by means of semi-structured interviews, also in both the EU and the USA. This will involve interviewing stakeholders from relevant companies involved in the development and application of AI as well as stakeholders from research, politics and administration. 

Duration

6.2024-12.2024

Clients

Mozilla Foundation