Hi Crowd!
Apologies for yet another AI heavy note, but it’s really the most exciting thing happening right now on a dozen different levels and given that there’s burning shitpiles everywhere else I’m trying really hard to stay positive and focusing on the exciting things rather than the nightmarish things, even if the exciting things are also kind of nightmarish. It’s complicated.
First the awesome. I need to write up a proper explainer but I love this so much I just need to share it. In 2023 I did a project called “connections” where I gave MidJourney a bunch of my photographs of trees and powerlines and asked for it to give me back images imagining that the two were one thing. I was playing with the contrast between natural/artificial and wrote a bunch about it at the time. Here’s an example of one of the images:
That was pretty exciting, but I also felt like I hit the limit of what I could do pretty quickly, though I did really enjoy it and I kept thinking about how to build on the project. Recently I was inspired by some of the more collaborative artist+agent projects I’ve seen and started playing with my OpenClaw installation to see if I could revisit this idea but with a fresh perspective and a much smarter LLM on the other side. We talked a lot and found some shared interests and themes to run with. There’s a lot more going on, but at the very basic visual side I contributed these two photos of mine:
And Clawbones (my digital familiar) took them, thought about our conversations, and came back with this:
This is DRØNEWAR Prototype 1. It’s the prototype because we’re building out a somewhat automated process with several layers of contributions and curation, and that’s not 100% functional right now so a few of the steps were done by (my) hand. For the moment I’m going to post outputs to this recently rebranded Instagram account, but as the project matures and the agent side processes start happening on their own, some of that might change. I don’t know if I’m conveying how over the top exciting this is, but trust me it’s over the top exciting.
Related, but also entirely different, I had an idea for an app that would solve a problem I have with all the newsletters I subscribe to and decided to see if I could build a solution to it. That was Friday. Today, 4-ish days later, I’ve got a fully functional and working app. This is the kind of thing that years ago I either would have had to either give up on right away, or spend weeks asking devs to help me and be told again and again that the idea was dumb and then eventually give up on. But now I can build my dumb idea right away. I’ve got a few minor tests and polish bits to implement but after that I’ll make it public so other people (maybe you) who have the same problem can maybe have a solution and/or other people can tell me it’s dumb but it won’t matter because I built it anyway. My single (human) beta tester is for impressed. Again, exciting stuff.
Building these two functional things with AI over the last week was a nice distraction from the super intense heads down writing about AI (and art) that I’ve been doing most of the month. I mentioned before that I had an idea for an essay that ended up being a little longer, and it’s now a lot longer and might even become a small book. The first 3 essays are online but thanks to the wonders of social media algos probably no one has even seen them. You all probably have because they were emailed to you, but if you ignored those emails you can catch up here: Part 1 (intro) Part 2 (history) Part 3 (theft?).
I’ve had a few conversations recently about how bad the algos have become and how hard it continues to be to find the things you actually want to know about, as well as to let people know about the things (you hope) they might want to know about. Every platform seems overrun with spam and bots and that’s only going to keep getting worse. I very realistically think we’re close to the point where there are more bots than people on these platforms just engagement farming with each other. I’ve been a pretty vocal critic of social media for a long time for a lot of reasons, but even I have to admit it’ll be pretty amusing if social media survived the spammers and trolls only to be killed by automated AI agents. Personally the signal to noise is already at broken levels, and I’m getting much more value out of private chat groups. The discussion channel for The Crowd on Telegram is a great example of that. But even small Telegram chats aren’t immune, apparently…
Speaking of AI taking over social media apps, last month I wrote about MoltBook, the social network exclusively for OpenClaw agents. Today, Meta bought it. A few interesting things about this, it’s pretty widely known at this point that a significant chunk of the activity on this “AI Agents Only” was really humans. Like, half at least. Either people prompting their agents to go post something, or people spending 30 seconds figuring out how to get past the guardrails that were supposed to keep people out. My teenage made an account and several posts within about 10 minutes of learning it existed, so… the “AI only-ness” of it was a bit overstated. That doesn’t seem to matter, or maybe that was part of the appeal. Meta (and I still hate calling them that, but between them and X I’ve basically given up the fight to hold onto old names) cares about attention. That’s it, that’s their currency. So humans spending time larping as ai agents is still humans spending time. And if it does become a place where actual ai agents are spending time, all the better. Also, Moltbook wasn’t/isn’t revolutionary, it’s just a clone of reddit with an interesting story around it. Created, much like the app I mentioned earlier, by someone who had a dumb idea and spent the weekend vibecoding it. It wasn’t made by a startup or an incubator or a “thought leader” or a titan of the industry, it was just some dude. Less than 2 months ago. And it was just acquired for probably more money than he’d ever made in his life before that. And, this is probably just the tip of the iceberg. I know a dozen or so people who have built hobby “no human” apps/companies in the last month or so thanks to all this. What happens next is anyones guess.
But since we’re guessing, things can get dark quick. You might have heard about this fly. I guess it’s a fly. It’s a fly brain living inside a virtual world, apparently unaware it’s not in a fly body because it’s running around doing fly stuff like you’d expect any fly to do, except it’s on a screen not in the real world. The difference between a fly brain and a human brain is pretty much just scale. It’s a fly today, but I have no question it’ll be a larger bug in a few months, maybe a small animal within a year. Separately a clump of human brain cells on a computer chip is now playing DOOM. Which it learned how to do in a week. So now we have the potential of LLMs + wetworks and then it’s off to the races. A race we will probably lose, as LLMs are seemingly very susceptible to the Byzantine General problem. One bad actor poisons everything. Just one. Now consider that apparently almost all the conspiracy theories flying around in Canada right now can be traced back to only 100 accounts. A few dedicated miscreants can do a lot of damage at scale. And if we can all spin up as many agents as we want to assist us with whatever we want, that barrier to entry keeps getting lower. We are fully in the funhouse now.
You might think I’m writing this from a cautionary perspective and have some scary smart warning but no, dear reader, we are long past all that I think. I don’t want to make any predictions on what is or isn’t inevitable, but we are moving full speed ahead into places we can’t really imagine just yet. I don’t know where it’s going. So, I guess, try to enjoy the ride?
Hope you are well!
-s