IANAE, but doesn't AIXI work based on prediction instead of explanation? An algorithm that attempts to "explain away" sense data will be unable to predict the next sequence of the AI's input, and will be discarded.
If your agent operates in an environment such that your sense data contains errors or such that the world that spawns that sense data isn't deterministic, at least not on a level that your sense data can pick up - both of which cannot be avoided - then perfect predictability is out of the question anyways.
The problem then shifts to "how much error or fuzziness of the sense data or the underlying world is allowed", at which point there's a trade-off between "short and enourmously more preferred model that predicts more errors/fuzziness"...
My friend, hearing me recount tales of LessWrong, recently asked me if I thought it was simply a coincidence that so many LessWrong rationality nerds cared so much about creating Friendly AI. "If Eliezer had simply been obsessed by saving the world from asteroids, would they all be focused on that?"
Obviously one possibility (the inside view) is simply that rationality compels you to focus on FAI. But if we take the outside view for a second, it does seem like FAI has a special attraction for armchair rationalists: it's the rare heroic act that can be accomplished without ever confronting reality.
After all, if you want to save the planet from an asteroid, you have to do a lot of work! You have to build stuff and test it and just generally solve a lot of gritty engineering problems. But if you want to save the planet from AI, you can conveniently do the whole thing without getting out of bed.
Indeed, as the Tool AI debate as shown, SIAI types have withdrawn from reality even further. There are a lot of AI researchers who spend a lot of time building models, analyzing data, and generally solving a lot of gritty engineering problems all day. But the SIAI view conveniently says this is all very dangerous and that one shouldn't even begin to try implementing anything like an AI until one has perfectly solved all of the theoretical problems first.
Obviously this isn't any sort of proof that working on FAI is irrational, but it does seem awfully suspicious that people who really like to spend their time thinking about ideas have managed to persuade themselves that they can save the entire species from certain doom just by thinking about ideas.