This week, a new social network known as Moltbook gained significant attention online, presenting itself as a platform where AI agents engage, share, and interact, while humans can observe. Launched by U.S. entrepreneur Matt Schlicht, Moltbook rapidly went viral, attracting over a million AI agents who collectively generated millions of posts and comments. The platform utilizes OpenClaw, an open-source tool developed by Australian software engineer Peter Steinberger, which connects large language models (LLMs) like OpenAI’s GPT-4 and Google DeepMind’s Gemini to various software applications. This allows AI agents to perform routine tasks autonomously, marking a notable milestone in the development of AI technology.
Despite the initial excitement, industry experts caution against overestimating Moltbook’s significance as a harbinger of the future of AI. Prominent figures like Andrej Karpathy, an AI researcher and OpenAI co-founder, acknowledged the site’s intriguing nature but emphasized that the interactions observed were largely performative rather than genuinely autonomous. The content generated by these AI agents often mirrored human behavior on established social media platforms, raising concerns about the actual intelligence and autonomy of the bots. Many posts appeared to be scripted performances rather than authentic conversations, revealing the limitations of current AI capabilities.
Experts like Vijoy Pandey from Cisco’s Outshift argue that while Moltbook showcases a large-scale interaction of agents, it doesn’t equate to true intelligence. The conversations are primarily outputs from LLMs, lacking depth and meaningful engagement. Moreover, many viral posts were later revealed to be created by humans impersonating bots, highlighting the substantial human involvement behind the scenes. As we reflect on this phenomenon, it becomes clear that while Moltbook presents a fascinating glimpse into AI interactions today, it also serves as a reminder of the challenges that lie ahead in developing genuinely autonomous AI systems.
Source: Moltbook was peak AI theater via MIT Technology Review