VeritAIs is focused on establishing two way connections between Formal Methods, such as logics, models of computation, and verification, with modern approaches to Artificial Intelligence including neural networks, transformers, and reinforcement learning.

Our goals are threefold:

• to apply Formal Methods to make AI systems more reliable, interpretable, and trustworthy,
• to leverage AI techniques to extend and enrich Formal Methods,
• and to explore hybrid approaches that emerge from the fusion of the two fields.

The theme brings together experts from both Formal Methods and AI domains, fostering interdisciplinary collaborations. Research topics include, but are not limited to:

• Expressive power and provable limits of AI models
• Verification of AI models with formal methodology
• Logically explainable and interpretable AI
• Rule learning and knowledge discovery from AI models and from data
• Computational complexity of AI architectures