Artificial Intelligence Efficiency & Brain-Inspired Computing - Artificial intelligence currently consumes 20,000 W of power, while the human brain only consumes 20 W for similar operations, highlighting the inefficiency of current AI systems [1] - The industry is studying the brain to transfer its mechanisms to artificial intelligence, using a scientific method of hypothesis testing based on animal behavior [2] - Engineers use differential equations and neuronal models to study emergent properties, which are then tested in animals to refine the models [3][4] Reward Systems & Learning in AI - Current artificial intelligence lacks a reward system, which is present even in phylogenetically distant species through dopamine receptors [5] - Dopamine release in the ventral tegmental area reinforces connections between stimuli and predicted rewards, accelerating learning and increasing the signal-to-noise ratio [8][9][10] - Experiments with rats showed that neuronal activity in the frontal cortex anticipates rewards as dopamine release intensifies, improving aspects of neural network learning [11][12] Perception & Consciousness Studies - Perception is subjective and crucial for survival, guiding behavior towards rewards or away from threats [13][14] - Studies on epileptic patients with implanted electrodes allow for in-vivo study of the human brain and consciousness during seizures [15] - Measuring integrated information in the brain correlates with consciousness scores measured by doctors during epileptic seizures [18]