Suppose you have a simple learner that can predict any computable sequence of symbols with some probability at least as good as random guessing. Then I can create a simple sequence that your predictor will get wrong 100% of the time. My program runs a copy of your program and outputs something different from your guess.
All the empirical evidence supports this. Good compressors have a lot of code to handle lots of special cases. On Fri, Aug 9, 2019, 8:15 PM Ben Goertzel <[email protected]> wrote: > > > >> >> Legg proved there is no such thing as a simple, universal learner. So we >> can stop looking for one. >> > > > To be clear, these algorithmic information theory results don't show there > is no such thing as a simple learner that is universal in our physical > universe... > > I'm not saying there necessarily is one, just pointing out that the math > is not so practically applicable as your statement implies... > > *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + delivery > options <https://agi.topicbox.com/groups/agi/subscription> Permalink > <https://agi.topicbox.com/groups/agi/T1ff21f8b11c8c9ae-M33ef00edadc8d2913f090b4f> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T1ff21f8b11c8c9ae-M8269e585c24ec57005dafb93 Delivery options: https://agi.topicbox.com/groups/agi/subscription
