@CartoonsHateHer I was bit by a dog off leash in a park almost a year ago, the owner saying it was just a puppy. I didn't notice it had broken skin, and was pascal's mugged into going to the hospital and getting rabies vaccine shots for the next couple weeks. (I hadn't asked for contact info)
@AvikonHadoop @Yoshua_Bengio@EpochAIResearch For what it's worth, there are a number of additional plots available on Epoch's site! https://t.co/5bm687Q3nh
@j_bollenbacher Hi John, I work with AE Studio who put out this paper https://t.co/PN4xILiczn you referenced. Want to chat with our team? We're very interested in this research for alignment + mitigating risks of suffering wrt AI consciousness.
In our new work, we show that Self-Other Overlap (SOO) training reduces deceptive behavior in models
SOO is a method that aligns an AI’s representations of itself and others
We think SOO may be promising for alignment, inspired by mechanisms fostering human prosociality
We are happy to release the first results of a RCT of a US program that provided $1,000/month unconditionally for 3 years to 1,000 individuals in the treatment group, with a group of 2,000 people receiving $50/month serving as the control.
These are sizable transfers. 1/ 🧵
Will disagreement about AI sentience lead to war?
AIs could become sentient and worthy of moral concern (similar to humans). But people will disagree about it, potentially causing conflict. In a new post series, I explore the dynamics and risks of the upcoming AI rights debate🧵
Be sure to catch #SXSW "The Path to Conscious AI" panel discussion with AE Studio CEO @juddrosenblatt along with @anilkseth, @allisondman and Michael Graziano. Sat, 3/9 at 2:30pm
https://t.co/CaVYGBbXZZ