OK. I've seen a lot of people here say that Eliezer's idea of a 'Bayesian intelligence' won't work or is stupid, or is very different from how the brain works. Those familiar with the machine intelligence literature will know that, in fact, hierarchical Bayesian methods (or approximations to them) are the state of the art in machine learning, and recent research suggests they very closely model the workings of the cerebral cortex. For instance, refer to the book "Data Mining: Concepts and Techniques, 3rd edition" (by Han and Kamber) and the 2013 review "Whatever next? Predictive brains, situated agents, and the future of cognitive science" by Andy Clark: http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=8918803

The latter article has a huge number of references to relevant machine learning and cognitive science work. The field is far far larger and more active than many people here imagine.