On AI Eschatology, Or The Trainwreck, Or Living in the Apocalypse

I’m lost like a lot of you. I don’t know how to live in the apocalypse. The biggest difference between me and most people is that I think the apocalypse isn’t coming; it’s already happened. 

Part One: The Trainwreck (2008)

I met Luke for the first time at a bar called the Backdoor at Roxy’s. You enter the Backdoor through maroon, velvet curtains. Then the curtains part, and you are suddenly in a prohibition-era lounge, with marble statues, bartenders that wear vests, vintage chandeliers, bookcases filled with liquor, velvet booths, and craft cocktails. I couldn’t tell you how their cocktails taste –I only drink Ranier tallboys and shots at the Backdoor. It’s a counter-culture thing.

Luke and I met because nearly everyone who knew both Luke and me demanded that we become friends. I couldn’t start a conversation about philosophy without someone saying, “Do you know Luke? You should meet!”

And so we did. That night we wound our way through avenues of thought and found that time and time again our ideas connected and amplified in conversation. We agreed on virtue ethics, trash-talked Plato’s theory of metals, and we were both befuddled and in love with twentieth-century phenomenologists. We were two new comrades in the quest to serve our lady Philosophy. 

That evening, we saw the Trainwreck for the first time. The Trainwreck is philosophy problem so catastrophic that any philosophy conversation I’ve had since has been an attempt to stop the chaos that the Trainwreck has caused, is causing, and will always cause. 

The Trainwreck is this: Given that we have a limited view of the world, how can we reason about what is true? We’re locked inside these mortal bodies, with minds that mysteriously arise from meat. How could such a being know what is true?

That night, we were two young philosophers in a speakeasy, chain-smoking and sipping whiskey and bad beer, and we had discovered the problem to end all problems. I feel like I never left that bar. 

You see it don’t you. You see that the further you chase this problem, the more structures break, the more the train tumbles off the track. If you can’t trust your own reasoning, how can you reason that you can’t trust your reasoning? What would it even mean to have a truth-oriented mind? Could any mind be untethered from this problem, or is this a feature of any observer? 

The problems wash into every corner of our assumptions about metaphysics, epistemology, and ethics like a dam bursting into a serene valley. It’s not just an unsolvable problem, but it’s a problem that destroys the problem-solving mechanism if you try to solve it.

Thankfully, in addition to shared philosophical intuitions and love of Ranier and whiskey, Luke and I also share ambition and hubris. If someone else hasn’t solved the Train Wreck, we would.

What’s more surprising is that we did it. First try. Almost like we’d been made to have that conversation. We covered every napkin we could get our hands on, but when the bartender announced last call, we had the first formulation of an answer.

If we can’t trust our own minds directly, could we create a distinctly “other” mind in order to better understand and improve our minds?

In everyday life, we use other people to check our own thinking, to ensure that we’re reasoning correctly and parsing reality accurately. Would an artificial mind provide an even better mirror into the limitations and mechanics of our own limited view? Could we see the errors and flaws in our own design by having a new mind to reflect back to us what it saw? 

Could these two separate types of minds bootstrap back and forth into a solution to the trainwreck?

We stumbled out of Roxy’s at 3 am, three sheets to the wind, but triumphant! And we thought we had solved the trainwreck.

We didn’t know that we’d started the apocalypse and there was no turning back.