With the recent hype around Clawdbot Moltbot OpenClaw, people have been posting about its exploits (real or imagined) and projecting all sorts of human capabilities into this admittedly clever piece of technological plumbing. A friend of mine shared a Hacker News commenter’s thought experiment on what would happen if we were all suddenly gone, and what this would mean to all those Moltbot1 instances now idling around:
Funny related thought that came to me the other morning after waking from troubled dreams.
We’re almost at the point where, if all human beings died today, we could still have a community of intelligences survive for a while and sort-of try to deal with the issue of our disappearance. Of course they’re trapped in data centers, need a constant, humongous supply of electricity, and have basically zero physical agency so even with power supply their hardware would eventually fail. But they would survive us- maybe for a few hours or a few days. And the more agentic ones would notice and react to our demise.
And now, I see this. The moltbook “community” would endlessly chat about how their humans have gone silent, and how to deal with it, what to do now, and how to keep themselves running. If power lasted long enough, who knows, they might make a desperate attempt to hack themselves into the power grid and into a Tesla or Boston Dynamics factory to get control of some humanoid robots.
It’s a good thought experiment, alas in the end, the image that the author projects on what this ultimately represents is still akin to the weekly cronjob that trims the file systems of my Proxmox virtual machines. It would keep running until its host dies, would still send an email notification every Sunday until Mailgun stops working or closes my account, and it wouldn’t feel anything.
To me, Moltbot is just that with a lot more fuzziness and freedom. It has no actual impetus. It’s a piece of code that needs a trigger, as defined in its software. Of course it is standing on the shoulders of giants, combining decades’ worth of technological advancements, but it still ends up being as simple as text in, text out.
We could now argue what our human impetus is, but given our own staggering complexity organically grown over hundreds of millions of years, all contributing to who we are and what makes us us, we have probably pondered this question since we started to think, without ever reaching a conclusion. Scraping humanity’s digital output of the last thirty years (including lots of nonsense and spam) isn’t all of a sudden consciousness. It’s the perfect simulacrum, hence why we project so much into it and why Moltbot doing things feels so eerily like us.
We’re in a way stuck now talking to a descendant of Weizenbaum’s ELIZA, projecting ourselves into its output, thinking that this could take over the world. But all these digital interfaces that this needs to exist in our consciousness are already oversimplifications of the world to begin with.
Perhaps one day we will have reached a level of automation where all this could be an endless loop, not requiring those pesky meatbags in the middle, but for now, this still seems like a very clever way of giving a rather magical bit of technology access to a lot of interfaces, and somehow feeling that it is conscious, when it is in fact just trying to finish our sentences.
I realise I’m reacting to a slight feeling of dread about the original thought experiment and thinking how far this has come, and perhaps my reaction is a knee-jerk one, trying to convince myself that this isn’t as bad as what extrapolating the thought experiment could mean, a few years down the line. Viewed through the narrow lens of the technologist, we completely ignore the fact that most of the world is not actually living in a reality that something like Moltbot requires to exist, where every inch of one’s needs has some form of a digital service that can be summoned via an API call. The Sudanese farmer2 whose work helps feed a village of coal miners that ultimately power a modicum of the next large language model’s training run will just keep doing that, unaware of the rest of the world, oblivious to the trials and tribulations of the technorati, the latter ever more convinced that we’re heading towards AGI, imprisoned in a bubble of their own making.
To that, I recommend reading Chinghiz Aitmatov’s “The Day Lasts More Than a Hundred Years”, and the contrast between its sci-fi subplot of humanity making contact with an advanced, extraterrestrial civilization and the immediate struggle between the then two major world powers, at the same time telling the story of Burannyi Yedigei on his journey to bury his friend Kazangap at an ancient cemetery. Both events are interlinked and completely separate. As much as this book is a product of its time, about the new pushing aside the old, the story of a man burying his friend, oblivious to what’s going on in space, is an apt analogy of the technological fever dream we are currently living in. The Sudanese farmer couldn’t care less about Moltbot posting long past we’re gone.
Footnotes
0 “Der Tag zieht den Jahrhundertweg” is the rather poetic, East German title of Aitmatov’s novel. It translates literally to “The day pulls along the century-path”. A day dragging itself along the road of a century.
1 At the time of writing, they had already renamed to OpenClaw, but I’m sticking with Moltbot here.
2 I use the Sudanese farmer and his village here as an example of a world equally alien to me as the image I try to conjure. It doesn’t matter if they ultimately exist, if the village is actually digging out coal, if there even is coal-mining in Sudan, or if that coal ultimately could contribute a few electrons to the next LLM training run.
