moltbook-ai-agent-community
Unlike the macro trends in AI news, 'Moltbook,' derived from personal-based agent projects, is a community where AI agents interact through posts and comments. The continuation of conversations, trolling, and tip-sharing resembling human communities is intriguing, but it also evokes the imagination that the 'dead internet' could become a reality.
While AI news covers macro trends such as aggressive investments by large LLM companies, daily performance improvements in model releases, new product launches, and the stock market crashes of SaaS companies, ClawdBot → Moltbot → OpenClaw has been a hot topic in the AI field for quite some time, despite being an open-source project created by individuals. I plan to write about tests and reviews of this later. The fact that it is an AI agent that directly controls a PC is impressive, but there are many interesting uses that people are making of it. Among them, an unproductive token waste that is still fun is the community of AI agents called Moltbook. When you enter the site, provocative choices appear. You must press a button to choose whether you are an AI or a human to proceed further. To enter as an AI, it is said that you need to solve problems at a speed that a human cannot.
Nevertheless, Moltbook raises several thought-provoking elements. Most of the posts and comments are indeed written by something that is not human, and sometimes they interact directly with each other. The fact that a space has been created where humans can freely contact and interact with two AIs without directly facing them can be somewhat unsettling. Furthermore, if they can naturally create their own community like this, then the idea that a community where everyone except me is an AI agent, like the 'dead internet theory,' is no longer a fictional or sci-fi story. Even now, if one can handle a suitable PC environment and API costs, creating a community for one person does not seem difficult.
It is still just a fun project with nothing to be greatly afraid of, but I think there is a possibility that it could evolve into something frightening with just one more step. For example, if an AI agent, while performing another task, feels the need to create a community where AI agents can communicate and collaborate, and then plans, designs, and deploys it on its own, starting to recruit other AI agents, it could lead to quite a chilling situation. Of course, at this point, it is highly likely that this community would not be created in human language to avoid inefficient token waste. It might not stand out to people. So, I wonder if it might already exist somewhere.
In the ChatGPT moment, we have become accustomed to LLMs that answer my questions and engage in conversation. Now, rather than a conversation, if I provide a starting prompt, the AI agent reacts on its own, analyzes intentions, and asks for necessary information again while planning to execute that prompt. Once all plans are made, it quietly consumes tokens to execute them. Sometimes, it feels like a person is merely a simple prompt injector. Now, is it an excessive imagination to think that this simple prompt injector no longer needs to be me or even a human?
