Not ever being much of a fan of Goertzel's arguments about AGI, it took some doing to get me to bother reading his April 14, 2020 paper "Grounding Occam's Razor in a Formal Theory of Simplicity <https://arxiv.org/pdf/2004.05269.pdf >". I only did so because while looking for yesterday's Joasch Bach's keynote address to Goertzel's AGI conference, I fell into a live video feed that, at that particular moment, was of a researcher talking my language about Solomonoff Induction. It turned out to be Arthur Franz of occam.com.ua, with whom I had communicated a couple of years ago. So I kept watching.
As they solicited questions from the viewing audience I asked a couple of questions about Solomonoff Induction. Not entirely to my surprise, these questions annoyed the panel (except Franz). The question I annoyed them with was posed in response to an assertion by Alexey Potapov to the effect that Solomonoff Induction could be biased to any prior one wished by one's choice of Turing machine. So I asked if there might be a notion of Turing machine complexity. This triggered Potapov, who shut down the topic by referring to Goertzel's aforelinked paper as "The Answer". Well, OK. So now where are we? Let's take this quote from Goertzel: "However, all these applications of Occam’s Razor either rely on very specialized formalizations of the 'simplicity' concept (e.g. shortest program length), or neglect to define simplicity at all." The notion that a universal computation is a "very specialized formalization" is _exactly_ the kind of Oracular nonsense that has narrowed my not filter on Goertzel over the years. Can someone bother to read this paper and report back? I can't be bothered given my other PRIORities. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T37756381803ac879-Ma686ba608b53c36b0af228c2 Delivery options: https://agi.topicbox.com/groups/agi/subscription
