Takeoff

[@MasamuneX]

i trust you because you heave a beard.

[@MasamuneX]

however some points to consider. We are currently assuming that all the gaps between the though will flow from one to the other in an intuitive way where as that isnt always the case when solving hard opened ended problems. Also hate to say it but "super intelligence" is more like here a a entity that has mental access to the tools needed to solve problems however I think there are generally only a few ways to solve hard open ended problems such as reducing the complexity inverting the problem and solving its opposite or the death by 1000 optimizations technique. Generally I think ai will be a soft SSI in the sense that it wont be able to introduce new methods of solving hard opened ended problems in a meta cognitive way it will probably just get stuck in the human rut of problem solving that has been established in its training data but do so at a level that is beyond us. I think ASI will be like a 1000 year experience software engineer butttt when you ask it to beat the stock market it will just use established problem solving techniques and not deviate from the basics of logical formalisms and understanding complex systems. On the surface laymen will be shocked because it suggests advanced things but under the hood will it truly innovate in the metacoganative sense?

[@gptboss]

yea pretty much my expectation too - but what the doc is saying was that it can synthetically produce a new training set that is more intelligent than the current one, ie solve problems at the highest level, then theoretically go beyond. We will see if it's an accurate prediction over the course of this year.