LEAM - Large European AI Models
Building large AI models in Europe. Open Source. Value-based. Universal.
What is LEAM?
LEAM is an initiative of the German AI Association (KI Bundesverband), and leading representatives from industry and research that fosters the development of large AI models.
To realize large AI models Made in Europe the initiative proposes six key steps:
- Collection and creation of comprehensive training datasets
- Support of excellent research in the field of AI
- Provision of hyperscale infrastructure
- Development of organizational structures and processes to establish a continuous workflow of model development and enhancement
- Integration of the models into the European innovation ecosystem
- Development of methods, benchmark datasets and criteria to ensure ethical requirements and European values
We want the models developed under this initiative to be open source and free to access for all market players. All European languages should be fully integrated into the models.
Why we have started LEAM
Recent progress made by U.S. and Chinese companies in developing large-scale AI models has the potential to transform the AI market. These larger models will outperform and gradually replace other AI solutions in the market. We fear that European actors will be left behind and the innovative European AI ecosystem will become dependent on American or Chinese solutions.
We expect complications with the support of languages other than English and we worry that American or Chinese AI solutions will not fully protect the privacy of its users.
To ensure that Europe stays competitive, we urgently have to invest in developing large AI models and AI supercomputers to train them. If we hesitate, we risk falling out of the fast AI innovation cycles.
What we want to achieve with LEAM
With LEAM, we aim to establish a European AI ecosystem that takes a competitive position in AI development.
We want our AI models to reflect European values and high quality standards in terms of data protection, transparency and bias, and want to make the training of AI models more sustainable.
Only with a strong AI ecosystem can we ensure the digital sovereignty of Europe.