Yeah, definitely both learning/reflection and preference-cascading are happening. Lots of people are learning about AI safety all the time these days. Then you have preference cascades happening on top of that, where all of a sudden it’s cool and normal (vs previously cringe and weird) to talk about AI x-risk in various new contexts/groups, like mainstream news. This increased visibility unlocked by the preference cascades (eg very many NYT front-page headlines recently) and endorsement by various elites/etc, then means that many more people learn about the issue and a greater fraction are convinced upon hearing about it. And so on.
Probably for the vast majority of the people involved, it feels like they’re just learning about a new issue (or maybe something they’d only barely heard of before), rather than changing their minds or finally speaking up about an issue in some dramatic way. So in that sense it’s mostly about learning, not a true preference cascade where everyone eg flips from espousing support for some leader to opposing them. But on the other hand, I think that preference-cascade dynamics are still very useful here for trying to understand things like “why is this only happening now, and why is it happening so fast”, or exactly which dominoes (high-profile people, organizations, etc) will fall next, or how long the cascade will continue before it fizzles out / meets opposing forces / etc. And, while the majority of people just feel like they’re “just learning about a new issue”, probably among the most important / influential people in society, disproportionately more of them are experiencing “changing their mind” or “speaking up for the first time” or etc (though many are still “just learning”).
Yeah, definitely both learning/reflection and preference-cascading are happening. Lots of people are learning about AI safety all the time these days. Then you have preference cascades happening on top of that, where all of a sudden it’s cool and normal (vs previously cringe and weird) to talk about AI x-risk in various new contexts/groups, like mainstream news. This increased visibility unlocked by the preference cascades (eg very many NYT front-page headlines recently) and endorsement by various elites/etc, then means that many more people learn about the issue and a greater fraction are convinced upon hearing about it. And so on.
Probably for the vast majority of the people involved, it feels like they’re just learning about a new issue (or maybe something they’d only barely heard of before), rather than changing their minds or finally speaking up about an issue in some dramatic way. So in that sense it’s mostly about learning, not a true preference cascade where everyone eg flips from espousing support for some leader to opposing them. But on the other hand, I think that preference-cascade dynamics are still very useful here for trying to understand things like “why is this only happening now, and why is it happening so fast”, or exactly which dominoes (high-profile people, organizations, etc) will fall next, or how long the cascade will continue before it fizzles out / meets opposing forces / etc. And, while the majority of people just feel like they’re “just learning about a new issue”, probably among the most important / influential people in society, disproportionately more of them are experiencing “changing their mind” or “speaking up for the first time” or etc (though many are still “just learning”).