The stated core goal of MIRI/the old SIAI is to develop friendly AI. With regards to that goal, the sequences are advertising.
Kinda… more specifically, a big part of what they are is an attempt at insurance against the possibility that there exists someone out there (probably young) with more innate potential for FAI research than EY himself possesses but who never finds out about FAI research at all.
A big part of the purpose of the Sequences is to kill likely mistakes and missteps from smart people trying to think about AI. ‘Friendly AI’ is a sufficiently difficult problem that it may be more urgent to raise the sanity waterline, filter for technical and philosophical insight, and amplify that insight (e.g., through CFAR), than to merely inform academia that AI is risky. Given people’s tendencies to leap on the first solution that pops into their head, indulge in anthropomorphism and optimism, and become inoculated to arguments that don’t fully persuade them on the first go, there’s a case to be made for improving people’s epistemic rationality, and honing the MIRI arguments more carefully, before diving into outreach.
You miss my meaning. The stated core goal of MIRI/the old SIAI is to develop friendly AI. With regards to that goal, the sequences are advertising.
With regards to their core goal, the sequences matter if 1. they lead to people donating to MIRI 2. they lead to people working on friendly AI.
I view point 1 as advertising, and I think research papers are obviously better than the sequences for point 2.
Kinda… more specifically, a big part of what they are is an attempt at insurance against the possibility that there exists someone out there (probably young) with more innate potential for FAI research than EY himself possesses but who never finds out about FAI research at all.
A big part of the purpose of the Sequences is to kill likely mistakes and missteps from smart people trying to think about AI. ‘Friendly AI’ is a sufficiently difficult problem that it may be more urgent to raise the sanity waterline, filter for technical and philosophical insight, and amplify that insight (e.g., through CFAR), than to merely inform academia that AI is risky. Given people’s tendencies to leap on the first solution that pops into their head, indulge in anthropomorphism and optimism, and become inoculated to arguments that don’t fully persuade them on the first go, there’s a case to be made for improving people’s epistemic rationality, and honing the MIRI arguments more carefully, before diving into outreach.