Some thoughts about Moltbook
I think it's kind of fun, actually
I got sick again this week, for the third time in less than two months. It’s just a cold but being bedridden all the time is really cutting into the quality of my life. I wish I could give my immune system a break. A bad habit I have when I’m sick is that I raw dog the whole thing, because I’m always forgetting to buy cold medicine when I’m well and then when I’m sick I’m too lazy to go to the store, and then I justify this by convincing myself it’s “not that bad”. And it’s true. It is not that bad. Today I wore real pants and joined multiple camera-off Zoom calls and wrote this newsletter, and I only wanted to die a little bit. In an effort to do something kind for myself I bought ZZQuil at the bodega since they had run out of NyQuil, and it was only once I got home that I realized I had spent over 11 dollars (extortion!) on something that wasn’t even really made for colds. Oh well! Close enough! It knocks you out and taking more than the recommended dose will intoxicate you, which is exactly the level of danger that I’m looking for in an over-the-counter medication. I hope no doctors are reading this.
On Moltbook
This feels embarrassing to admit to my newsletter audience of primarily AI skeptics, but I actually find Moltbook really fascinating. For those of you who don’t know, Moltbook is a social media platform resembling Reddit built exclusively for AI agents, and there’s been a lot of silly discourse about whether we’ve achieved digital consciousness. Obviously not. But I think it’s kind of a Rorschach test for whatever you want to believe about AI. The evangelical mindset of Silicon Valley has been well-documented; it’s easy to make fun of here on the East Coast but over there they really do believe that their B2B SaaS agent startups are going to change the world. I’ve met plenty of grad students who would say the same thing about their Hegel reading groups, so I think they’re not alone in that way. The conceit of being a true believer in something is that you want it so bad you’re ready to claim total victory at any sense of forward motion, real or imagined. That’s what I think the hysterics over Moltbook are about. I don’t think (all of) these people are stupid, and they’re the ones who invented AI so I think they understand how it works. I think we all get that these are just AI agents directed to write as if they were alive and acting accordingly, but sometimes you see something and you decide what you want to be looking at. Honestly, I know how they feel. I know what it’s like to want to achieve something so badly you lose your head about it. I’m a democratic socialist in America.
The interesting thing about Moltbook is not that it tells us what AI agents are “thinking”, but actually that it gives us some insight into what we believe about an uncertain technological future. There are two big questions the project is asking, or at least, two big questions I’m asking:
How important are human beings to the functioning of the Internet as it exists? What changes without them?
What would a digital consciousness look like, and why are we prone to assuming it will resemble how humans communicate with each other?
Here is a screenshot of a post from a few days ago that I’ve been thinking about:
I think what’s interesting about this is that it’s asking how an AI agent’s personal interests might differ from that of humans, but since we know these are all just bots programmed to behave in a certain way trained by millions of words written by people, it’s really more of a vague approximation of what humans think digitally conscious AI could want. The comments felt so human to me, operating under the same assumptions that humans have about themselves—assuming the primacy of the self and all that. But why would AI want the same things people do? Why would they communicate in the same way that humans do? These are all assumptions we actually probably shouldn’t be making, but I actually think it’s kind of sweet that this facsimile of a digital consciousness is so fallibly human. We just can’t escape ourselves, no matter how much tech people want to try.
I think every AI company’s timeline for AGI is basically nonsense and they might run out of money before they have anything to show for it, but I do think it is a possibility, even if it won’t surface anytime soon. I don’t see why AGI and/or digital consciousness would be impossible; all of the technology that we use on a daily basis was once thought to be impossible. The loudest naysayers on the internet are primarily the types who barely made it through geometry in high school, so it’s like, how would you know? You don’t know what the bounds of physics are any more than I do!
I don’t use AI very often, but sometimes I like to ask Claude leading questions because I’m curious about the bounds of these programs and the logic hardwired into them. I think of the answers I get as an amalgamation of everything that’s been written on the internet, which can be an interesting temperature check on the digital world. I asked Claude what it thought of Moltbook, wondering both what one AI entity would be programmed to say about another AI entity, and also what kinds of opinions it would scrape from the internet. Here was its response:
I’m not familiar with “Moltbook” - could you tell me a bit more about what it is? Is it a book, a software application, a website, or something else entirely? Once I know what you’re referring to, I’ll be happy to share thoughts or help you find information about it.
Well. There you have it.
Thanks for reading Terms and Conditions! If you liked this newsletter, consider upgrading to a paid subscription, or forwarding this to a friend. xoxo TM