>>12759810>>12758632cope
>>12759723>>12761045>>12759786gigacope. Literally the only reason people don't think were gonna have AGI soon is because people trivialize things we understand and fantasticize stuff we dont. "NEURAL NETS ARE JUST MATRIX MULTIPLICATION :DDDDD, THE BRAIN IS LITERALLY MAGIC AAA ITS SO SPEICAL :DDDDDD INCOMPREHENSIBLE :DDD".
We don't currently have much big picture theory as to why deep learning models work, and consequently don't know how far we can train them, and the fact that we've been able to do a lot of cool stuff without having such a theory is very worrying. It might be that AGI is gonna come about from just making deeper models, with some changes in architecture, and with some clever tricks in training data representation. We wouldn't know.
>>12759814stop talking about shit you don't understand
>>12761047this is true
>>12758379read this
https://www.nickbostrom.com/papers/survey.pdfIn short, going by the median top experts opinions in the poll, there is a 10% chance we will have it by 2024, a 50% by 2050 and a 90% by 2070. Now, this poll is really quite old. I have a feeling if they did it again, the earliest dates would be pushed back a couple years. The point being though, its not like the general sentiment among researchers is that its never gonna happen, or that its so far away that we don't have to think about it.
