The German association Large-Scale Artificial Intelligence Network (LAION) has launched a petition, calling on the European Union (and several other states) to establish a publicly funded and democratically governed research facility capable of building large-scale artificial intelligence models.
LAION is best known as the creator of LAION-5B, a training dataset for text-to-image models such as Stable Diffusion. As such, it is a key organization promoting an open-source, public interest driven approach to AI development.
The petition is particularly interesting in the light of the recent open letter calling for a moratorium on AI research (see our previous opinion). LAION researchers propose an alternative approach: the role of the states and public institutions should not be limited to regulating, and possibly curbing, commercial AI development. There is a need to develop alternatives that are publicly funded, driven by a public-interest mission, and overseen by democratically elected institutions.
The LAION proposal marks an important moment in the development of open source ecosystems. Since the summer of 2023, open source models such as the Stable Diffusion and BLOOM have become prominent examples of an approach that offers an alternative to closed, corporate development of AI technologies that runs the risk of centralizing power.
By proposing a mission to build a “CERN for AI”, the LAION researchers argue that a strong role for the public sector is needed to ensure public interest in this field. The growing role of the public sector in open source ecosystems also emerged in conversation with Mike Linksvayer during one of our Open Future sessions. The petition argues for a democratization of AI technologies in a double sense: 1) through democratic oversight of their development, and 2) the democratization of the technology itself through its open sourcing.
The responsible use of AI is a key issue in public debates about AI research and its regulation. There is an argument that the principle of responsible AI development is incompatible with open source models. Most notably, OpenAI –- the leading commercial developer of AI models – has pivoted away from a commitment to open source, ostensibly due to concerns about security and responsible use. The LAION proposal offers an alternative where responsible use is ensured in open source environments through the involvement of democratically elected institutions.
In this respect, the proposal comes at a crucial moment in the development of the European policy approach to open source AI systems. Right now, it’s focused on regulatory issues (as the proposed AI law is being amended to regulate general purpose AI, including open source systems). The petition launched by LAION points to the need for a broader policy focused on building public capacity. As Janet Haven of Data and Society argues in her recent opinion, these can be seen as securing not only digital policy goals, but also a broader civil rights agenda.
At Open Future, we have been advocating for European policies that support digital public infrastructures through public funding of key digital technologies, developed and shared in open source ecosystems. And more generally, we have been arguing that public institutions need to be empowered to become key players in digital ecosystems, if we want them to function as digital public spaces. The LAION proposal applies this general vision to the field of AI development. And shows that Europe – as the AI Act is being finalized – should do more than just regulate the development of commercial AI systems.