A deep dive into the new Manson Family—a Yudkowsky-pilled vegan trans-humanist AI doomsday cult—as well as what it tells us about the vibe shift since the MAGA and e/acc alliance's victory
Impressive writing, and great historical analogy-making. I've been following the Ziz thing since it broke, and you're the first person I've seen actually cite and quote their blog. Hopefully reading it didn't inflict any permanent harm on you. Also really appreciated the deeper connections you were able to draw to Manson family via the EA vs e/acc conflict. Great work!
Thanks Vincent. Great analysis! It's extremely interesting all of this. I think Yann Lecun - he was interviewed by Lex Fridman recently - convinced me that the AI doomers are incorrect in their bleak prophecy of AI catastrophe because they don't understand the tech properly. A friend of mine who works in the industry keeps threatening to leave his job because he disagrees so fundamentally with them calling it AI. It isn't AI: chatgpt etc are just large language models. They aren't intelligent like an embodied creature. They can generate sentences based on probability algorithms - that's it. They can play chess/go/etc really well because they're closed systems that can be algorithmically broken down. There's no way that alphago or chatgpt could conceive of humanity as a threat or anything like that. And these applications can't even do something as simply embodied as playing pool yet. The computing power required for that - still a closed system - isn't available yet. They are still not even very good at recognising and manipulating images yet. Nowhere near as good as a young child. Love the course!
Thanks! Hard agree that ChatGPT and other LLMs aren't going to terminate humanity or even achieve general intelligence any time soon—and primarily for the reason you mention that they are probability algorithms (inferring from past datasets that inevitably are confronted with the problem of induction). Not so much here but in other writing, I'm more focused on shifting conceptions of AGI to include things like markets among other complex self-organizing multi-agent systems. When all the doomsday rhetoric around AGI is seen more as an account of capital than of LLMs, I think it might be much more plausible. Even some of the doomers seem to be starting to see this. Here’s Max Tegmark on the Lex Fridman pod in 2023: "the issue with capitalism and the issue with runaway AI have kind of merged... Anyone who has concerns about late-stage capitalism having gone a little too far, you should worry about superintelligence because it’s the same villain in both cases."
Remembering Dennis Potter's 90s TV series Karaoke/Cold Lazarus which predicted a future dominated by the internet and virtual reality; the rebels couldn't do much but be terrorists to the web-people and spray their slogan RON (Reality Or Nothing) on walls.
Hadn’t heard of this show before. Makes sense it aired the same year the Unabomber was apprehended. No doubt we’ll get a Zizian inspired Netflix series soon enough. Certainly still looks like the only options for the neoluddites are RON or run.
Impressive writing, and great historical analogy-making. I've been following the Ziz thing since it broke, and you're the first person I've seen actually cite and quote their blog. Hopefully reading it didn't inflict any permanent harm on you. Also really appreciated the deeper connections you were able to draw to Manson family via the EA vs e/acc conflict. Great work!
Very nice of you to say, I’m glad you enjoyed!
10/10, seriously high quality. Thank you Mr Lê.
🙏
Thanks Vincent. Great analysis! It's extremely interesting all of this. I think Yann Lecun - he was interviewed by Lex Fridman recently - convinced me that the AI doomers are incorrect in their bleak prophecy of AI catastrophe because they don't understand the tech properly. A friend of mine who works in the industry keeps threatening to leave his job because he disagrees so fundamentally with them calling it AI. It isn't AI: chatgpt etc are just large language models. They aren't intelligent like an embodied creature. They can generate sentences based on probability algorithms - that's it. They can play chess/go/etc really well because they're closed systems that can be algorithmically broken down. There's no way that alphago or chatgpt could conceive of humanity as a threat or anything like that. And these applications can't even do something as simply embodied as playing pool yet. The computing power required for that - still a closed system - isn't available yet. They are still not even very good at recognising and manipulating images yet. Nowhere near as good as a young child. Love the course!
Thanks! Hard agree that ChatGPT and other LLMs aren't going to terminate humanity or even achieve general intelligence any time soon—and primarily for the reason you mention that they are probability algorithms (inferring from past datasets that inevitably are confronted with the problem of induction). Not so much here but in other writing, I'm more focused on shifting conceptions of AGI to include things like markets among other complex self-organizing multi-agent systems. When all the doomsday rhetoric around AGI is seen more as an account of capital than of LLMs, I think it might be much more plausible. Even some of the doomers seem to be starting to see this. Here’s Max Tegmark on the Lex Fridman pod in 2023: "the issue with capitalism and the issue with runaway AI have kind of merged... Anyone who has concerns about late-stage capitalism having gone a little too far, you should worry about superintelligence because it’s the same villain in both cases."
Ah, yes, in those terms, the outlook seems fairly bleak. I'm reading Yanis Varoufakis on technofeudalism at the moment. I'll have a listen to Tegmark.
Remembering Dennis Potter's 90s TV series Karaoke/Cold Lazarus which predicted a future dominated by the internet and virtual reality; the rebels couldn't do much but be terrorists to the web-people and spray their slogan RON (Reality Or Nothing) on walls.
Hadn’t heard of this show before. Makes sense it aired the same year the Unabomber was apprehended. No doubt we’ll get a Zizian inspired Netflix series soon enough. Certainly still looks like the only options for the neoluddites are RON or run.
Excellent summary, thanks for all the work you put into these two pieces
Transon Family.
e/-ACK