"As simple as possible but no simpler" (I believe Einstein's paraphrase of Occam's Razor), means not so simple that the theory disagrees with observation.
Occam's Razor is true because any possible probability distribution over a set of strings (descriptions, theories, programs) must favor shorter ones than longer. For any string, the set of less likely strings is infinite, but only a finite subset of them could be shorter. Remember that the goal of AGI is to do everything that a human can do (with the obvious application of automating labor). Machine learning is a big part of that, and Occam's Razor is an important guiding principle. But ultimately it comes down to experimentation when deciding which languages to use to encode strings. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T37756381803ac879-M6e9a0b22514355703d69a762 Delivery options: https://agi.topicbox.com/groups/agi/subscription
