FAIR

Well beyond currently available technologies, FAIR aims at devising and developing AI systems capable of interacting and collaborating with humans, of perceiving and acting within evolving contexts, of being aware of their own limitations and able to adapt to new situations, and interact appropriately in complex social settings, of being aware of their perimeters of security and trust, and of being attentive to the environmental and social impact that their implementation and execution may entail.

In short, FAIR aims at building an AI that does not yet exist. To meet this level of ambition, FAIR envisions new forms of AI that are:

1. human-centered: co-evolve with the human “in-the-loop”, at individual and collective level
2. integrative: bridge across the different AI methods, technologies, disciplines, and competences
3. resilient: operate in challenging, noisy, uncertain real-world settings, 4. adaptive: perceive, learn and act in dynamically changing environments
4. high-quality: meet quality standards for high-risk, high-accuracy, safety-critical applications
5. symbiotic: promote effective human-machine interaction and collaboration
6. edge/exascale: operate at the infinitely small on the edge, and at the infinitely big on the cloud
7. pervasive: operate ubiquitously in different social settings
8. green-aware: consider the environmental dimension “by-design”
9. sustainable and bio-cognitive: mimic the principles of biological systems at multiple scales