Justin Riddle #11 - Bayesian Inference & Quantum Computers

In episode 11 of the Quantum Consciousness series, Justin Riddle pitches a novel model of quantum computers as undergoing Bayesian learning to generate a model of how the world works and to guide future behavior. To get started, we talk about the act of observation in the double-slit experiment. When you look at which slit the photon traveled through then the wave function was destroyed and the photon behaved like a particle. The act of observing the wave function collapsed it! But what is observation? Is this merely the extraction of information? Roger Penrose would suggest that the atoms that compose the measuring device interacted with the photon and that interaction collapsed the wave function. It doesn't matter if "you," the experimenter, saw the result of the observation. The example of Schrodinger's cat further emphasizes the confusion that arises when we treat superpositions as knowledge and not as reality. What then is knowledge? How do we infer the future? One solution comes from Bayesianism, which provides an update rule for how we can test our theories on how the world works by collecting data and updating our theories based on this data. Artificial Intelligence has made great use of this simple update rule to create a predictive model of future events and to make an action plan for what to do next. Interestingly, Quantum Bayesianism proposes that the probability distributions of wave functions can be modeled using Bayesian principles. At the end of the video, we speculate that the cycling between digital computation and quantum computation might implement some form of Bayesian learning whereby digital information is acquired and then an internal model of the world is updated in the quantum computation phase. With each collapse of the wave function, the quantum computer could learn from the environment and then plan its future actions.