New theory of/ roadmap to AGI: novelty detection/ seeking


hi all, just joined. not easy to find this forum from the openAI home pg!

recently, inspired by alphazero and its conquest of both go and chess using the same reinforcement learning algorithm, decided to finally write up some old but bold ideas on AGI that are in line with these new experiments.

comments welcome! also would be very interested to contact/ work with any developers who are looking for some cutting edge design/ architecture ideas/ leadership.

this site seems like nearly 1-of-a-kind on the internet for AI enthusiasts to hang out/ discuss/ work on prjs. have others heard of any other places? there seem to be some links on the openAI site (older blogs) to gitter pages, but they seem to now come up 404.

ps heres other essays on AI developments from last few yrs.

thx to @elon for funding the openAI initiative, think its a great idea & hope it continues to innovate, prosper and break new ground.