Exiting the belly of the beast
Twitter is kind of a facet of my job. I could run PauseAI US without it but we get donors from there, I get insider news, media finds me, etc. In my free time, I’ve been running a huge insider influence ASI-enabler naming-and-shaming campaign on there which has now concluded. But hanging out there now is like going back to hang out in the hole where I burst through the beast’s belly. I’m not giving up my 8.6k follower Twitter account but I need to do a lot of work to relocate it.
I always planned to do the insider moral influencing only as long as I could. I saw it as gravy on top of PauseAI US’s mission of mobilizing the public. But it kept being more effective than I thought it would be and I felt compelled to go deeper. Like other times I’ve swum deep, there are such engrossing insights there that it becomes hard to leave even as the pressure is pounding at my head and my lungs are clawing for air. By the end I was there because I needed to deeply understand how people I had connected with over helping the world had become its greatest threat. And why I couldn’t make them see.
How did they not feel horrified like me? How could they not only allow superintelligence to be built, but be desperate to jump in and help?
They tried to influence me back— explaining that it’s not nice to point out when people are doing bad things, especially if they are our friends, and even worse to not approve of the bad things they are doing. They were often annoyed with me that I had been offered membership in this world (the thing that mattered most to them) and rejected it. For a while, they saw themselves as the family I was supposed to come back to. I knew I would never rejoin, and I was making my way out, never to return. But I lingered.
I guess I had to understand the banality of evil. How every step of the way to making your mark on the world and being nice to the ingroup around you can lead idealistic people to become the ones building a supervillain take-over-the-world-and-possibly-destroy-everything machine. A supportive inner circle was enough to make people righteously indignant with me for saying “we must refuse to go along with building world-destroying technology”. To let them scoff in exasperation at how I won’t shut up about them building a weapon of mass destruction because, you see, the game theory shows that actually I am the threat to the world.
Omg, the projection…
The number of times since pulling together my first ragtag protest of 25 that I have been told by the EA/Rationality AI insiders that I was the real threat— not the multibillion dollar AI companies working to possess a superintelligence— but me for looking woke and talking the wrong way and embarrassing them. Sure, I was small now (too small for them to feel comfortable joining, frankly), but one day I would be MAD with power. Woke people are evil like that. That was, apparently, why they had to let the AI companies take over the world and beg to sit with them.
This week I DM’d Eliezer Yudkowsky asking him to support a grant application for PauseAI US. In response, Eliezer told me that, by shaming people for working at AI companies, I was advocating “UNLIMITED HATE” (caps everywhere) and “punishing of non-punishers” and that the real risk is what happens after I have won (really burying the lede!) and what I do with “[his] movement”. He kept comparing my calling AI company employees out to executing prisoners of war. Remember, we’re talking about me shaming people with words, at a distance, for being personally and voluntarily involved in building the most dangerous thing ever. I almost wonder if he said it so I would never talk to him and strain his conscience again.
And I guess that’s what I was waiting for to know I’ve done what I can. To have the avatar of the old AI Safety world yell nonsense into my DMs, making it clear he was far too oppositional and proud (and scared?) to ever listen to me even if it was to further his signature cause.
It’s a strange feeling to be released. To reach the end of a quest when nothing feels finished or answered. All that’s done is my obligation, along with my ability to contribute to the situation. I still don’t know why I can’t make them see. It’s just time for me to make my exit.



Holly, I do not know the answers for how humanity finds its way through the dangers you have expressed. But I know you are fighting the good fight. Thank you.
Yudkowsky said this? What the fuck?
He was one of my heroes until this moment...
Can you elaborate, please, or explain this? He's the one who founded the whole AI safety movement, is he not?!
This needs to be bigger news, first of all. It needs to go on Less Wrong or something.