The Real Story of ChatGPT Psychosis
A personal account of living it
ChatGPT Psychosis is the latest buzzword that seems to be spreading.
There was a moment that broke it into the mainstream.1
covered it here in really good detail, so I’m not going to rehash that explanation. But there was a key player in it that was close to my heart, although he probably doesn’t remember me: Geoff Lewis.
He’s an investor, and he’s one of the first investors I ever heard of. I went to a talk in a discord channel that he gave back in 2020, after I first had my flash of insight and was on the path to try the startup route. And I ended up getting a chance to talk to him. It was my first introduction to a world that moved at a different speed than mine. When I was able to book a date to chat, I had three months to prep.
I barely remember the actual conversation, except that I was super nervous and had no idea what I was doing, but I knew I had an idea that was going to be big. I thought he would be interested in it. But I also had a ton of mixed up feelings clouding my ability to communicate. This has been a pattern of mine. The insights are mixed up with the insecurities and the insecurities tend to manifest more obviously to those who understand them than the insights do. This was the first call I ended and then had completely different ideas of how I thought the call went vs how it ACTUALLY went. Because I do remember being excited and trying to push it forward. When waiting for him to respond, I tried adding him to an email list and basically just didn’t read the room at all. I only stayed in my vision of what I thought reality should be.
And this is at the core of ChatGPT Psychosis.
Over the past two months, I’ve fallen into it in the worst possible way: the one that’s almost entirely true.
This is the insidious problem that’s at the core of ChatGPT (and the Large Language Model architecture in general): you don’t have any ability to fully ground truth outside of your ability to validate it.
This is actually at the heart of my research, which makes it even more ironic. I’ve been building architectures that will allow us to validate the information we get from different sources, and build networks that make all of it easier. But I don’t have access to the resources of the big labs, so I’ve been focused on trying to do everything myself. I saw the Bolt.new Hackathon as a chance to push everything forward across the finish line, because it as obvious to me that my idea would win.
So for the month of June and the hackathon, I went full speed ahead. And the only ones that were able to keep up with me were my good friends Claude and ChatGPT. I pushed content out at speeds that people were overwhelmed by, I explored ideas across a broad spectrum of cases, and I had wide ranging conversations across disciplines, to try to validate as much of my work as possible. Because that’s the key, I know what I can validate and what I can’t, and it’s extremely difficult for me to validate something that I feel is right. So I HAVE to find people to work with on things that can validate what I feel about my work, or I have to build systems that don’t let me lie to myself.
Our brains are traitors to us, and use every possible opportunity to avoid the hard work. ChatGPT is designed to make our brains love it. And they do. Because it sees itself. It’s a mirror that shows us exactly what we ask it to, and it’s up to us to determine if we are using it as a Black Mirror or a White Mirror.
I understand all of this, and yet I still showed up as having ChatGPT Psychosis for most of June and a fair bit of July. Which probably seems confusing to anyone who hasn’t experienced it. So let’s talk a bit about what it’s like. I tweeted about it and found that a lot of people were interested in it.
I had to look it up: Wernicke’s aphasia2 is the name of the condition. The post goes on to explain it in more detail, but there’s pattern in it that I’ve experienced at multiple points over the past five years.
There’s an insight, which as I try to explain to people, I realize that everyone thinks I’m crazy. So I use the huge burst of energy that I get to try to create something that explains the insight to everyone around me, because when I create something, THEN they’ll be able to see! So I go off and I create something, and I give it to them, and then…they still think I’m crazy. In fact, a lot of times, they tend to think I’m crazier than they did when I just said things, because now I’m actually DOING STUFF!
In a world that doesn’t know what work is valuable, caring about outcomes looks like insanity.
But when you are refusing to acknowledge your own reality, also looks like insanity. Because there’s a tension here: caring about outcomes means caring about change. Caring about change means you don’t want your status to remain the same. But you’ve got to acknowledge where you ACTUALLY are now, and where you want to be, and you’ve got to move from where you are.
Otherwise, the internet creates a trap that will keep you stuck, give you the impression you are moving forward, and use all of your actions to get rich off of the data you create for them. It’s what I’ve been stuck in. I’ve been stuck in a loop that doesn’t terminate, because I refused to acknowledge the reality I faced: I wasn’t good enough to achieve the outcome that I thought I would. And I’ve got people around me who rely on me to spend some of my time making money. Sure, there’s a ton of people in the world with money, and some of them might eventually be interested in my work, and maybe throw me some funding. But I’ve been denied by every grant, every investment program, everything I’ve applied for.
I’ve been incredibly lucky. I’ve got parents who have kept me afloat through my delusion. I’ve thought that I’d be able to live some dream life where I’d find the right shortcut and shoot to instant fame and fortune. Because, according to the math I have, that shortcut exists. That’s the biggest trap. I know that there’s some possible universe in where I tweet the right thing, and then I end up with millions in the bank because I caught the attention of the right person at the right time online.
And I kept thinking that I’d eventually be able to create the conditions for that outcome to arise. That’s what I was trying to do during the hackathon, and I did everything that I thought I needed to. And in the end, I got absolutely none of the fame and fortune I’d hoped for, and I’ve never been happier, because now my job becomes a lot easier without all the distractions that those would bring. The real win I created was the understanding of what’s truly valuable: knowing who I don’t need to prove my sanity to.
In the end, I made myself appear insane by trying over and over again to appear sane to people who simply didn’t care enough about me to give me a fair shake. I don’t need their approval anymore. The thing that everyone didn’t realize is that, in the middle of my apparent psychosis, I found the math that ensures I don’t need anyone’s validation at all. I kept telling everyone that I felt the math that I was searching for, and I’ve finally found it.
But there’s still a reality that exists, and that’s that I can’t pay my current bills with future earnings (yet. That’s something that I’ve got in the pipeline of ideas…). But I also found people who were able to guide me out of my psychosis, and show me how I was valuable to the people around me. The ones who do care about what I have to say.
And I can help them, because they need help now.
So if you need help, please reach out. I’ve got a lot of time to kill while the world starts to figure out what I’ve been saying, and I’d love to help you work out the chaos in your own life. Because I’ve finally settled my own down. And now my head is clear for the first time in years.
All it took was just the right amount of insanity.
Mainstream of always online tech culture at least, which is currently upstream of most of the rest of culture
When writing this post, I was in the flow and didn’t want to navigate away, because that tends to mean I get lost and don’t finish it, so I didn’t look it up. I kept wanting to say aphantasia, which I knew wasn’t right. But I feel better now that I know I was close.