Both agents have the same complexity after training but not before. On Wed, Nov 21, 2018, 1:24 AM ducis <ducis...@126.com wrote:
> > > Forgive me for not understanding the Legg paper completely, but > how would you separate a 1MB "AI agent" executable plus a 1PB file of > trained model (by "sucking data from internet"), from a 1PB executable > compiled from manually built source code? > I don't see how the latter can be classified as complex while the former > being classified as simple. > > > -- > ----------------------------- > > 在 2018-11-20 01:15:10,"Matt Mahoney via AGI" <agi@agi.topicbox.com> 写道: > > > > On Mon, Nov 19, 2018, 11:12 AM ducis <ducis...@126.com wrote: > >> >> Hi Matt, >> Doesn't the "predictor" actually contains trained models as well? >> > > Yes. That is the normal way to write a predictor, like in a data > compressor. It collects statistics on past input to predict future input by > looking up the current context or guessing what pattern or program > generated the data. > > It doesn't change the fact that universal predictors don't exist. You can > always simulate any predictor and output the opposite bit. > > -- Matt Mahoney, mattmahone...@gmail.com > > *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/T9c94dabb0436859d-Md2b8109f555d8c3f1cc8dcc3> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T9c94dabb0436859d-M4f73704f2f705b359d4c7de2 Delivery options: https://agi.topicbox.com/groups/agi/subscription