What physics offers for artificial intelligence? Lessons from the brain's inner time and its dynamics

Area of Science:

• Neuroscience and Computational Theory
• Physics of Time and Dynamics
• Artificial Intelligence

Background:

• Physics provides a framework for understanding time and dynamics, crucial for activity patterns.
• The brain exhibits intrinsic neural dynamics and spontaneous activity, enabling dynamic input processing.
• Current computing paradigms, including AI, lack inherent temporal dynamics and 'inner time'.

Purpose of the Study:

• To explore how computing can leverage physics' concept of time and dynamics.
• To investigate the brain's use of intrinsic neural dynamics for processing and participating in physical time.
• To highlight the limitations of current computing and AI in engaging with real-world temporal dynamics.

Main Methods:

• Analysis of physics principles related to time and dynamics.
• Review of empirical evidence on brain's scale-free activity and variability as 'inner time'.
• Comparison of biological neural processing with classical and non-von Neumann computing architectures.

Main Results:

• The brain utilizes scale-free activity and variability ('inner time') to actively track and encode input dynamics.
• Neural activity, through entrainment, aligns with external rhythms, enabling participation in physical time.
• Current computing devices lack spontaneous activity and 'inner time', preventing active engagement with dynamic environments.

Conclusions:

• The brain's 'inner time' and dynamic repertoire are essential for active information processing and real-world participation.
• AI and current computing are 'locked out of time and world' due to their passive processing.
• Bridging this gap requires incorporating principles of dynamics and intrinsic temporal activity into future computing architectures.