After the last AI winter we were trained to always say "machine learning" and never "AI", because AI was considered a dirty word that would get your grant application instantly denied. As recently as 2017, even bringing up AGI could get you dirty looks at NeurIPS. I think that aversion to even speaking about AGI out loud has led the community to really underestimate the risk. I'm glad Dan Hendrycks is talking about this.
After the last AI winter we were trained to always say "machine learning" and never "AI", because AI was considered a dirty word that would get your grant application instantly denied. As recently as 2017, even bringing up AGI could get you dirty looks at NeurIPS. I think that aversion to even speaking about AGI out loud has led the community to really underestimate the risk. I'm glad Dan Hendrycks is talking about this.