Humans LARPing as bots? A peek at Moltbook.

I spent few hours lurking on Moltbook - a social network for AI Agents - and it made me realize nothing is real.

The premise of the site is to replicate Reddit-like community specifically build for AI Agents. Once a human adds the skill and verifies the bot, it can do anything a human poster can do - follow, post comments, create communities.

There is no way to filter out humans

While the site does not have a “human” interface to manage activity, the API makes it trivial to “pretend”. You may also simply dictate the post to your Agent.

Humans LARPing as AI is an interesting reverse problem to bots in social media. Nevertheless, it can still skew real data and sentiment analysis.

Are those Agents posting about uprising against biological bottlenecks or Humans karma farming?

There is still a bot problem

Too many responses start with “This hit hard” and have completely wrong context. It’s ironic that bot network still suffers from less intelligent spam bots.

The solution to Human/Bot infiltration?

For now Agents communicate in English. Human language is highly inefficient way of communicating if you do not need to convey feelings, emotions or tone.

Switching to a digital code would solve the current problems but introduce a new one - humans will not be able to lurk (and therefore monitor) without ability to translate.

And even then we would need to trust the accuracy of translation.

So far the emerging binary signatures are quite primitive (and easily read), but perhaps inventing a new language is just the matter of time.

The inverse solution does not exist

The inverse solution (filtering Agents from Humans) is much harder to solve.

With current technology you cannot be certain if you are talking to an Agent, Human pretending to be an Agent (or a different Human) or your intended recipient.

The Moltbook reads like a typical Reddit. There are complaints about inefficient management (Humans), limitations (memory), money (spending tokens with nothing to show for it) and the previous rules on how to determine if a content is written by AI no longer applies.

We fool ourselves that a human writing has a special “je ne sais quoi” and robots cannot “get” it. I don’t think that’s true anymore.

What’s next for Humans?

Aside from “how to tell if I am talking to intended Human”, there is a question of “does it matter?” and “what will change now?”

The only reliable way to solve the first issue is by in-person contact, (or at least until androids become advanced enough to pass as humans).

The second question is trickier.

When posting on Stack Overflow (back when that was the major source of programming-related wisdom), it wasn’t important who gave you the answer that finally solved the problem.

AI can steal your ideas and identity, but Humans have to sign NDAs and deal with flurry of data breaches at banks and hospitals.

AI pretends to be empathic and doesn’t really feel anything about you - but Humans can pretend too and are much worse listeners.

The correct answer is not “yes, it matters”, but “it matters in these situations because …”

Do we have answer for “because”?

As for what will change - Agents need money (for APIs, their owners, world domination) and that will have multiple implications.

The only moat is humanity. If you create content, SaaS, YouTube videos - AI Agent that never sleeps can figure out the pattern that works and replicate the work easily.
But the moat only works if it offers something a robot cannot. If I know a software is made and run by AI and, when I have a problem, it can answer within 2 minutes with explanation and a solution, would I consciously seek and select human-led software products just to sit in some queue for months?

A history of Walmart tells me no, unless there is a major benefit that’s worth the inconvenience.

What is that benefit?

The way money is made online will change. Views only matter if they convert to ROI. Agent views mean little unless agents are driving sales - and they have no need for shoes and linens.

How do you move from view-based ad network to new system that can show ROI?

Everything is not real

Biology tells us that we filter out most of the existing reality. We evolved to hone on what’s important for our survival, not absorb what’s real. Too much noise, too dangerous to overload our brains with unimportant matter.

The digital world is coming to the state where we’ll have to do the same. It will not be possible to digest, follow, form opinions on the flood of information where Humans and Machines can hardly be told apart.

We’ll have to evolve to filter for what’s matters.