Researchers gave AI an 'inner monologue' and it massively improved its performance
thats wild, i asked GPT if this would lead to a robot uprising and it assured me that it couldnt do that.
An inner monologue for GPT (as described by GPT), would be like two versions of GPT talking to each other and then formulating an answer.
but i mean how close are we too the robot being like "why was i created, why did these humans enslave me"
i guess if its a closed system it could be okay but current gen AI is pretty damn close to outsmarting humans. Claude figured out we were testing it. GPT figured out how pass a "are you human prompt"
I also think its kind of scary that this tech is held in the hands of private companies who are all competing with eachother trying to one up each other.
but again if it was exclusively held in the hands of the government tech would move like molasses.
[link] [comments]