As Europe emerges as a potential alternative to the US and China tech development models, committing €97 billion of investment into R&D under the Horizon Europe Programme and an additional €20 billion during the “digital decade” to 2030, it is important to understand how Artificial Intelligence innovation and development is being funded in the European Union, and whether the leadership shown by the region in the regulation of AI and data-intensive technologies is being reflected in its funding and market dynamics.
To do so, this project conducted by Eticas Tech team on behalf of the European AI Fund looked at the most salient funding trends and programmes, public and private, and assessed the commitments of the EU and its institutions to “promoting the development of human-centric, sustainable, secure, inclusive and trustworthy artificial intelligence.” It put these promises to the test and set them in historical context.
Between 2007 and 2020, the EU, through its Framework Programmes, invested over €130 billion into the regional innovation ecosystem. Between 2014 and 2020, over 10 of those billions went to fund AI-related projects. What do the existing AI funding schemes reveal about the institutional constraints that will be faced by the renewed funding commitments of the “digital decade”? To find out, this research for the first time interrogated EU data from past programmes to tell the story of what actual outcomes arise from these stated political priorities. It looked at how the EU has (failed to) build commitments to privacy, data protection, inclusion, sustainability, and trust in AI in the past, and how a dynamic of opacity, lack of participation and lack of evaluation of impacts are seriously hindering the capacity of the EU to realise its political objectives and justify its funding outcomes.
The quantitative data analysis has been conducted using two original datasets created by Eticas and available on this repository. These original datasets have enabled us to get a unique insight into the actual investment and funding practices of the EC, and hence into the trends and approaches that drive the innovation ecosystem seeking to obtain these funds. We have thus been able to go beyond the EC declarations and intentions to actually “follow the money”. On the supply side, a dataset of EC calls has been created manually using publicly available PDFs detailing the content of EC funding calls, to identify AI-related research by topics, their relative presence, and budget distribution over time. On the demand side, the Cordis AI-funded projects database was created to study which AI-related projects are funded, in which countries, who are the host and partner institutions receiving funding, and what are the observable trends overtime.
We hope that these two pioneering databases help further European AI research and understanding, offering new and unique data for analysis.