How I Became an AI Safety Person

I'm currently dedicating my life to AI safety: trying to lower the chance that AI leads to human extinction and other catastrophic harm. I think AI risk is the most important problem in the world, so I'm proud to be working on it.

Why am I so worried about AI?

I think there are a variety of persuasive reasons to care about AI safety. 

But this post is not primarily about the rational reasons to care about AI safety. Instead, it's about the personal story of how I, Jacob, started caring, and how it enveloped my life, and how I became an "AI safety person". Maybe, just maybe, you'll realize it's who you want to be, too.

As a child, the problems of the world always weighed on me. I remember crying when I learned about factory farming, and when I learned about climate change.

But I found the concept of artificial intelligence utterly boring. 

My dad was always the AI enthusiast in the family. Sure, I liked math, but not really science. When it came to Roomba robots or Amazon Alexa, I never cared at first — I was fine living the way I was. (These days, Alexa performs such vital services for me as "playing music in the shower" and "calling me Bartholomew because I told it once that my name was Bartholomew and I don't know how to change it.") My dad also loved to read books or watch movies about futuristic technology — Isaac Asimov novels, The Terminator, Star Trek (always The Original Series, with Captain Kirk and Mr. Spock) — and sometimes I liked them when he recommended them, but I never got obsessed. Plus, they seemed like science fiction.

~

These days when people worry about AI killing everyone, they often go to lesswrong.com. I first stumbled upon lesswrong.com in 2019! So you might think that I was an AI-worrying prodigy. 

However, I never really read the posts about AI in those days! Instead, I got hooked on reading posts about cognitive biases and game theory — authored by the world’s foremost AI doomsayer, a self-taught man with a cult following and the name of Eliezer Yudkowsky. 

But I didn't know anything about Yudkowsky's AI opinions at first. I just knew his writing was good (as I wrote on this very blog in 2019, "Eliezer Yudkowsky over on LessWrong is really cool. I wouldn't recommend it to everyone, as it's a bit technical and jargony, but it's definitely insightful and deep if you can get past that part. The link is right here.") — even his Harry Potter fanfiction (the first forty chapters, anyway), where I learned about the insane inflation of the Galleon. (Did you know that in Book 1 of Harry Potter, the Weasleys' Gringotts vault has only one Galleon, and Harry's wand costs seven Galleons, but by Book 4, Fred and George are buying a JOKE WAND for five Galleons.)

So I lived happily in ignorance. 

~

I almost wrote an article on this blog about AI alignment before ChatGPT even came out. I would've looked so prescient and cool! (I'm embarrassed that that was the first reason to come to mind … sorry that it wasn’t "Those of you who were reading the blog then could’ve been more informed about the future…")

I didn't write it then because I didn't feel I understand the risk enough to explain it. 

I think I had watched this Eliezer Yudkowsky video. And I'd listened to this CGP Grey podcast "Twenty Thousand Years of Torment." Grey had been a techno-optimist, but the video — a review of Nick Bostrom's book Superintelligence — opened his eyes to structural difficulties of building AI. 

I randomly picked up Superintelligence from the library. And then ChatGPT came out. 

(This post will hopefully also discuss how I got into AI safety after ChatGPT, my attempts to write this post, and a short overview of what I consider to be the persuasive argument for AI safety.)

Note: I'm running a blog-a-thon today, and I am therefore deontologically obligated to publish this post, but I don't consider it done. I thought the HIPPOCAMPUS one was satisfactory but this is always a topic I've struggled to write about; you win some, you lose some. I plan to make further edits to it in the future. Please bother me to do this.