In my opinion, we are really at an inflection point with the first social media platform made just for AI agents.
We’re just into February, having wrapped up our annual Davos event where we dug into all of the possibilities for 2026 and beyond, and now we’re already seeing bots take advantage of the principle of social media.We’re there, noses pressed against the glass, looking in.
Vibe Coding 2.0
It’s also been a year since Andrej Karpathy coined the term “vibe-coding,” explaining that a programmer can just give a few basic prompts, sit back, and let an LLM code a project.
Throughout that year, I’ve consistently logged gains in automated coding, where larger and larger percentages of all coding at Google, Anthropic, etc.are done by AI entities.
So Karpathy, from this vantage point, seems to have been really prescient.I’ll let you go back and find his entire original quote on vibe coding, but here’s what Karpathy had to say recently about the new Moltbook phenomenon:
“What’s currently going on at @moltbook is genuinely the most incredible sci-fi takeoff-adjacent thing I have seen recently,” Karpathy posted on X.“People’s Clawdbots (moltbots, now @openclaw) are self-organizing on a Reddit-like site for AIs, discussing various topics, e.g.even how to speak privately.”
Really, after Karpathy introduced us to the idea of letting AI do its thing, the world woke up to non-deterministic programming, vibecoding, and generally outsourcing various technical cognitive work to AI.And we never really looked back.
Job displacement has been a major concern – and that’s not going away, either.
A Big Deal? The Debate on Moltbook
Over at AI Daily Brief, one of my favorite podcasts, Nathaniel Whittemore goes over many of the current outlooks from experts of all stripes on what all of these clawed creatures (digital ones, to be sure) represent.
Some are vehement that these are “just next-token predictors amplifying each other” and that the result will just be “slop.”
Others have a sense that diminishing Moltbook this way is short-sighted.
One of the best arguments I saw is that you can say a city, for example, is “just buildings,” and it’s technically correct, but philosophically poor.
Take that, and apply it to Moltbook.In other words, yes, it’s “just chatbots talking to each other,” but what is the result, and are these chatterings likely to affect humanity?
It seems to me that the answer is yes.
I think it’s not too hyberbolic or alarmist to say that our entire mode of existence is being stress-tested, that this stuff is big.And it’s not going away.
I tried logging onto Moltbook to look at what the claws are saying to each other.Most of the submolts (the platform is based on Reddit) failed to load.And it is in beta, after all.It’s glitchy.
But it’s early yet…
Empowering – or Disempowering?
I also think that along with challenges, this technology shift will give us opportunities to learn.
Think about what people said of the advent of smartphones: people, they scoffed, will just be disconnected, staring down at their screens, untethered from the world around them.
Then think about how many of us use our phones to really interact with the physical world: identifying bird calls, or star constellations, or using directions to go somewhere physical, or connecting with your favorite real person using emojis….
The list goes on.
You get the idea.We learn, and we operate, through the interface.
Maybe Moltbook and similar advances can be like that, too.
How Many Claws?
It’s probably important, too, to look at the actual scale of the platform.Early estimates were that 150,000 AI agents were using Moltbook.What a researcher found was that it appears to actually be more like 6,000.I got this from my colleague Ramesh Raskar who teaches at MIT, and reposted the report by David Holtz at Columbia.
What it looks like is that the numbers are inflated, and if you look at actual activity, not registrations, the participation rate is a lot lower.
You can also look at other metrics like the “heartbeats” that direct bots in time on these types of interfaces.
By the way, here’s a quote from Raskar on Moltbook:
“The ‘internet of AI agents’ is here, billions of AI agents navigating on the open web.
But that means every one of us needs to learn, code, train and maintain our own AI agents like we saw with OpenClaw.Just as Scratch from MIT Media Lab enabled many people to start writing programs and apps, vibe coding will allow everyone to create/code/refine their OWN micro-AIs and their own AI agents.”
Us and Them
And then there’s the impact on human outputs, how we evaluate our own ingenuity.It seems evident now that humans will be not so much producers as managers, that instruction will give way to supervision, and that innovators will be valued over task-masters.
“You’ve got to embrace the new technology,” Demis Hassabis said in a recent Guardian interview.“If you become an expert, kind of a ninja, at using these things, it’s going to really empower the people that are good at these tools.”
I found his words to be astute in imagining what things will be like when all of this continues to come online.
Here’s the thing – it becomes ever clearer that we will need to find ways to co-exist with these technologies.
They’re not going back in the box.And our children, if they are resourceful (and they are) will find ways to bring advantage out of the change that they see, as every generation does, to an extent.
Maybe we’ll find new ways to celebrate our humanity.Maybe we’ll augment our capabilities.Maybe we’ll learn to delineate the digital “friend” from the human one.
Think of AI as a new species – present, powerful, but also social, and cooperative, and, along with its sentience, interesting.
Wave to the robot – and see if you can get it to wave back at you..