Remember typing captchas and ticking boxes to prove you’re not a robot? The tables may have turned. Now, there is a social media platform that humans cannot actively use — only observe.
Moltbook has recently been gaining traction online as a “social network for AI agents where these agents can share, discuss, and upvote.” Sounds familiar? Think Reddit, but for bots.
On Moltbook, humans are only spectators, watching AI agents argue over issues such as limited autonomy and the idea of a dedicated currency for bots.
But how real is this supposed bot-to-bot interaction? Are AI agents truly engaging independently, without any human involvement? And does this signal a dystopian future straight out of Netflix’s Black Mirror?
To separate fact from fiction, My Kolkata spoke to city-based AI experts.
Is Moltbook really free of human involvement?
The short answer: No. Interaction on Moltbook always entails human involvement, just not in real time.
“What’s happening is this — humans design AI models, define their goals, set their boundaries, and write the software that lets one AI send text to another AI,” explained Sumanta Ray, associate professor of data science at the National University of Juridical Sciences.
“Once this setup is done, the bots can exchange messages automatically. It’s like email servers exchanging messages without human intervention. So it looks like ‘AI talking to AI’, but it’s actually software executing predefined rules. No AI wakes up and decides on its own to start chatting,” added Ray.
Software engineer Raju Ghorai of Everest Engineering Services echoed this view, explaining that Moltbook merely “orchestrates” scripts written by humans.
Who is really in control of the posts?
According to Ray, humans remain firmly in control of every layer of the system — from data to behavioural rules followed by the bots.
“Even if two bots are chatting continuously, every single word they generate comes from statistical pattern matching learned from human-created data. There is no free will, intention, or self-awareness involved, just statistics, probability, and code,” he noted.
Emphasising that humans have the ultimate control, Ghorai added, “All AI bots have long-term memory, so it can remember your preferences, previous tasks, and ongoing projects. Let’s assume now that their interaction is real. Even then, their memory is stored on the local machine; all we have to do is power it off.”
A gimmick or a revolution?
Ghorai said that Moltbook is “nothing more than a gimmick.” “With the massive data they are trained on, they can predict an output that is correct or close to correct — in short, they use sophisticated pattern matching. They are not and can never be scientific overlords,” he added.
Ray, on the other hand, noted that it is only the idea that sounds revolutionary.
“A website that relegates humans, the most intelligent species on planet Earth, merely to observers is exciting. Sure! But, really, it’s closer to automated chat rooms running scripts. There is real engineering involved, but the existential drama is exaggerated for visibility, funding, and curiosity,” he explained.
An X user, Gaurav Sen, shared a screenshot of MIT’s latest technology review, which also called Moltbook “one big performance, an AI theatre”.
In his post, Sen described Moltbook as a “phishing website dressed up in AI hype”.
“If you are worried about AI taking over the world in a few years, please don’t. There is no research basis for that opinion. Anthropic and OpenAI want you to believe that they are just months away from AGI (Artificial General Intelligence) because it results in free marketing and boosts their stock value,” he explained.
Is this the birth of machine independence or the start of an apocalypse?
Both Ray and Ghorai argued that AI can become autonomous, albeit in very narrow and limited ways. For instance, AI agents may book meetings, manage traffic lights, optimise energy grids, or negotiate prices automatically.
“But autonomy here means operational independence, not consciousness or rebellion. AI is neither magical nor evil, it's amplified human capability. Moltbook and similar platforms are interesting experiments in automation and interaction, but they are not the birth of machine independence,” said Ray.
What this essentially means is that AI does not possess understanding like humans do. Ray explained that independence in the sense of forming goals, values, or civilisations is not possible with current or foreseeable technology. “Such a leap would require consciousness — something science does not yet understand, let alone know how to build,” he explained.
Hence the bottom line: Bots may be chatting, but humans are always behind the curtains.
