Our focus is on energy-efficient architectures inspired by biological neural systems, as well as non-neural AI paradigms such as symbolic approaches or hybrid systems. Other topics include AI models inspired by principles of quantum mechanics, automated reasoning (e.g. multi-scale reasoning, mathematical proofs), working with sparse or unconventional data formats, and architectures that can better handle long dependencies and temporally continuous data streams. Also of interest is the development of generative models that learn by simulating environments, as well as adaptive systems that can integrate new information in real time.

The focus is on fundamentally redesigning the way machines learn. This includes approaches such as federated learning, which ensures data protection and scalability, as well as alternatives to gradient-based learning. Intrinsically motivated mechanisms for reinforcement learning, causal learning paradigms, active learning and building synthetic data ecosystems for robust AI development are also supported. The aim is to find new ways of making machines more intelligent by using and combining data sources more efficiently and by developing new data modalities.

Promoting innovations in hardware that enable a new type of AI. This includes technologies such as analog computing, reservoir computing, neuromorphic computing, photonic computing and holographic computing. Differentiable logic gate networks and biohybrid systems that integrate living tissue with robotics and AI are also in focus. Robotic systems with embodied cognition and adaptive learning that can interact better with their environment through tactile and haptic feedback will also be developed. The aim is to create AI systems that demonstrate real-world intelligence through their physical and sensory integration.

Supporting unconventional, visionary ideas that go beyond established thought patterns. These include approaches such as meta-learning, evolutionary algorithms, self-organizing and self-improving AI systems and neuromodulation-inspired AI. The simulation of real systems, including physical, economic and social dynamics, as well as the development of artificial life forms that mimic emergent intelligence behaviors are also being promoted. Also of interest are digital organisms with life-like behavior patterns, cooperative AI approaches, swarm intelligence and emotional intelligence in AI to enable nuanced human interactions. Self-aware and self-reflective intelligence are also among the innovative concepts that can be considered.