AI system self-organizes to develop features of brains of complex organisms

Cambridge scientists have discovered that applying physical constraints to an artificially-intelligent system can enable it to develop brain-like features in order to solve tasks. The researchers created a simplified version of the brain using computational nodes and observed that the system developed characteristics and tactics similar to those found in human brains. The physical constraint applied to the system mimicked the organization of neurons in the human brain, where communication between neurons becomes more difficult the further apart they are. The system showed similar problem-solving abilities and coding flexibility as human brains, suggesting that physical constraints play a role in shaping brain organization. The researchers believe that understanding these constraints could shed light on differences between individuals’ brains and contribute to the study of cognitive and mental health difficulties. The findings also have implications for designing more efficient AI systems, particularly in situations where physical constraints are present. They suggest that AI systems tackling similar problems to humans may benefit from brain-like architectures. For example, robots operating in the physical world with limited energy resources could benefit from brain structures similar to ours. The research was funded by various organizations including the Medical Research Council, Gates Cambridge, and Google DeepMind.



Source link