After the open letter calling for a moratorium on AI development, I publish an exploration of various perspectives on AI existential risk that breaks down the debate into four logical groups, and considerations of the insights that each group provides.
Apr 5, 2023·edited Apr 5, 2023Liked by Chase Hasbrouck
I've been following this stuff for some time now and I think this is an interesting and useful piece of work. It will be interesting to see how this evolves.
One further comment. It seems to me that belief in AI x-risk exists mostly along a Silicon Valley to London/Oxford axis. That suggests to me that x-risk has a cultural aspect to it. Here's a post where I wonder why the Japanese don't seem worried: https://new-savanna.blogspot.com/2022/05/whos-losing-sleep-at-prospect-of-ais.html
Bill, thanks for reading and appreciate the praise.
Definitely agree that there's a cultural aspect to this. I came to the party late so I'm not familiar with how it started, but it feels like there's a nexus between EA-flavored longtermism, rationalism of the LW variety, and strong AGI existentialism.
Love the idea of how cultures can play into this, too. Ghost in the Shell is the prototypical transhumanist AI experience.
Yes, Chase, those three are definitely connected. My guess the core is a half-dozen to a dozen or so individuals in the Silicon-Valley /London-Oxford axis counts for a lot of that.
I love Ghost in the Shell. Notice that there's no rogue AI in it. Unless you count the Puppet Master, but just what is THAT? And whatever it is, it's not trying to eliminate humanity. Of course, when the Wachowskis got a hold of it and transformed it into The Matrix, things changed quite a bit.
You might want to take a look at a book, Inside the Robot Kingdom, by Frederik Schodt. It's a cultural history of robots in Japan. Schodt is a translator who's done a lot of work with Osamu Tezuka (Astro Boy). He also did the English translation of the Ghost in the Shell manga.
This is cool, thanks for sharing. We're starting to see people think through and apply recursive principles, vs. wave one of dev (which was just ChatGPT clones and other wrapper-type things).
Thanks for the very handy and useful categorization of the positions. I hope you keep going in this analysis. I'm in the futurist's camp. I still stand by my analysis, "The Myth of the Superhuman AI" published in Wired way back in 2017. Still valid. https://www.wired.com/2017/04/the-myth-of-a-superhuman-ai/
Kevin, thank you for the kind words. I actually had your interview with Noah Smith in mind when I first started drafting the piece. I really should have included you and Kurzweil on the chart, actually.
Absolutely. There's a spectrum of beliefs here, and people will move one way or the other. The biggest group is the one I didn't mention, actually, and that's the "no opinion" public. As that group activates and sorts, you'll see changes. (Also when this issue gets politicized, though I haven't seen much of that so far.)
I feel like the futurist utopian thing is a bit of a stretch. I think there's probably some people like me that think AI risk talk is ridiculous (the 'misinformation' and 'ethics' risks and the x-risks) but also that AI isn't going to be wildly transformative (at least not overnight).
I really despise the x-risk and the 'ethicists' though, because it seems like a pure political play to take control over an emerging technology.
I've been following this stuff for some time now and I think this is an interesting and useful piece of work. It will be interesting to see how this evolves.
One further comment. It seems to me that belief in AI x-risk exists mostly along a Silicon Valley to London/Oxford axis. That suggests to me that x-risk has a cultural aspect to it. Here's a post where I wonder why the Japanese don't seem worried: https://new-savanna.blogspot.com/2022/05/whos-losing-sleep-at-prospect-of-ais.html
More generally, I've been posting on the subject for awhile. This is a link to my posts: https://new-savanna.blogspot.com/search/label/Rogue-AI
Bill, thanks for reading and appreciate the praise.
Definitely agree that there's a cultural aspect to this. I came to the party late so I'm not familiar with how it started, but it feels like there's a nexus between EA-flavored longtermism, rationalism of the LW variety, and strong AGI existentialism.
Love the idea of how cultures can play into this, too. Ghost in the Shell is the prototypical transhumanist AI experience.
Yes, Chase, those three are definitely connected. My guess the core is a half-dozen to a dozen or so individuals in the Silicon-Valley /London-Oxford axis counts for a lot of that.
I love Ghost in the Shell. Notice that there's no rogue AI in it. Unless you count the Puppet Master, but just what is THAT? And whatever it is, it's not trying to eliminate humanity. Of course, when the Wachowskis got a hold of it and transformed it into The Matrix, things changed quite a bit.
You might want to take a look at a book, Inside the Robot Kingdom, by Frederik Schodt. It's a cultural history of robots in Japan. Schodt is a translator who's done a lot of work with Osamu Tezuka (Astro Boy). He also did the English translation of the Ghost in the Shell manga.
Not another book for my reading list! :) Thanks for the rec.
Appreciate the graphic and categorization.
Just wanted to share that it may not even be the big labs creating the risk now that so much is public.
AutoGPT seems to be a step along the paperclip scenario.
https://github.com/Torantulino/Auto-GPT
This is cool, thanks for sharing. We're starting to see people think through and apply recursive principles, vs. wave one of dev (which was just ChatGPT clones and other wrapper-type things).
Thanks for the very handy and useful categorization of the positions. I hope you keep going in this analysis. I'm in the futurist's camp. I still stand by my analysis, "The Myth of the Superhuman AI" published in Wired way back in 2017. Still valid. https://www.wired.com/2017/04/the-myth-of-a-superhuman-ai/
Kevin, thank you for the kind words. I actually had your interview with Noah Smith in mind when I first started drafting the piece. I really should have included you and Kurzweil on the chart, actually.
https://noahpinion.substack.com/p/interview-kevin-kelly-editor-author
(Long-time fan, by the way. I adored WIRED as a kid and have been a Cool Tools reader for years.)
Do you foresee the -ists transferring amongst groups as AGI develops further?
Absolutely. There's a spectrum of beliefs here, and people will move one way or the other. The biggest group is the one I didn't mention, actually, and that's the "no opinion" public. As that group activates and sorts, you'll see changes. (Also when this issue gets politicized, though I haven't seen much of that so far.)
Which of these 4 groups would you say would be most closely associated the label “Foomer”?
I feel like the futurist utopian thing is a bit of a stretch. I think there's probably some people like me that think AI risk talk is ridiculous (the 'misinformation' and 'ethics' risks and the x-risks) but also that AI isn't going to be wildly transformative (at least not overnight).
I really despise the x-risk and the 'ethicists' though, because it seems like a pure political play to take control over an emerging technology.
Sounds like you're a Pragmatist, then! :)