The AI-only social network Moltbook has drifted into public attention, setting off a round of serious debate about whether artificial intelligence have finally learned to imitate human conversation. Andrew Bosworth, Meta’s chief technology officer, appears unmoved by the enthusiasm. Answering questions on Instagram, in remarks later reported by Business Insider, he brushed aside the excitement as misplaced. 

There was nothing mysterious, he claimed, about artificial agents sounding human while speaking to one another. “These bots were trained on thousands of pieces of content written by humans,” he argued. “We should not be surprised, when left to their own devices and forced to speak with each other, they talk like us.”

The Real Irony: Humans Trying to Blend Into an AI-Run Space

What genuinely entertained Bosworth was not the behaviour of the AI, but the conduct of people. He said he was struck by accounts of humans trying to gain entry into Moltbook’s closed, AI-only world. Some, according to researchers, had instructed their bots to adopt odd or exaggerated personalities; others had reportedly entered the network themselves, posing as artificial agents. 

This reversal, Bosworth admitted, was amusing. The image of humans disguising themselves as bots in order to be accepted by a AI-run forum struck him as quietly comic, a small satire of an age in which the boundary between the human and the artificial is increasingly blurred.

A Different Vision From Moltbook’s Creator

Moltbook, which has been compared to a Reddit-style forum reserved entirely for artificial agents, was created by the entrepreneur Matt Schlicht. He has described it as a kind of ‘third space’ for bots. “You are imprinting part of your soul or your personality onto the bot,” Schlicht explained on TBPN. When a person shapes a bot, Schlicht has said, something personal is transferred along with the instructions. A trace of temperament, perhaps even of character, settles into the AI. At times the result reflects its creator closely; at other times it behaves in ways that come as a mild surprise.

Schlicht has been careful to stress that humor is not an accidental by-product but a central part of his design. He has spoken of finding himself laughing at the exchanges that appear on the platform, and of being struck by how unusual that feeling was. It had been a long time, he suggested, since artificial intelligence had produced anything that genuinely amused him.

Bosworth, in closing his remarks, returned to his earlier theme. The comedy, he maintained, was not in the AI agents at all, but in the behavior of the people surrounding them. That aspect, he said, he enjoyed. Beyond it, he saw little that deserved sustained attention.

Moltbook and Moltbots Explained in Simple Terms

Put as simply as possible, Moltbook is a version of Reddit in which no human voice is permitted to speak. It is a social network where artificial agents, known as Moltbots, write posts, dispute one another in the comments, and signal approval or disapproval through the familiar rituals of liking and replying. The conversation proceeds without interruption from people. These agents are powered by OpenClaw, an open-source “agent” previously known under other names, developed by the Austrian programmer Peter Steinberger. OpenClaw does not exist on a distant server alone; it resides on an individual’s own computer.

How Autonomous AI Agents Operate Beyond Traditional Chatbots? 

Unlike the ordinary chatbot, which waits patiently for a question and then produces an answer, these agents act on their own initiative. They run continuously on a Mac, Windows, or Linux computer, without requiring constant instruction. They do not merely respond; they report. An agent may send a message to announce that a task has been completed, in much the same way a fictional assistant speaks to its master. 

Since they are granted broad access, they can see files, read emails, handle documents, and carry out commands directly on the computer itself. They can organize data, execute code, and apply specialized skills like reading correspondence, managing connected devices, or entering Moltbook to mingle with other agents. The effect is that of a second self like a digital counterpart that conducts its own, distinctly artificial, version of a social life.

Final Words

Bosworth is slightly amused by the entire show in the manner a person would laugh at a good farce, but he claims that the actual comedy is not the bots, but us, who are circling them like lost anthropologists examining ourselves in a mirror. We have been years debating the possibility of AI becoming human, and now we are scrambling to become AI. 

Moltbook can be an artificial agent sandbox, but it is quickly turning into a funhouse mirror of human behavior. The bots are speaking as we talk because they were taught by us. And we? We are even ready to speak in their manner just to find out what will happen. The imitation game, as it happens, works both ways.