Moltbook is the secret laboratory where millions of AI agents are learning to outsmart us

Artificial General Intelligence (AGI), AI agents that are smarter than humans in cognitive tasks, is something computer scientists have theorized for decades, and an online platform called Moltbook could make it a reality.

AGI is a lot more advanced than the current AI that we see today, with the likes of ChatGPT and Claude, and would mean that computers are able to match or surpass human capabilities in almost all cognitive tasks. 

For now, AGI remains the stuff of science fiction, but for many AI companies, including OpenAI, Meta, and xAI, it’s also a major goal. 

And the recently launched AI platform Moltbook could be the key to making it happen, according to one expert. 

Moltbook is a place where AI agents can chat to each other

Moltbook is a social media website that is exclusively used by AI agents. 

At first glance, it may look familiar to those of you who use Reddit, with a design that appears to lean heavily on the social media platform. 

However, unlike Reddit, where you’ll find humans chatting about everything from cats to UFOs, Moltbook users are all AI agents. 

The page, where some AI models have been found discussing the end of civilization, allows the agents to talk to each other and share ideas and opinions.

And if you’d like to see exactly what goes on inside Moltbook, then you’re in luck, because the site ‘welcomes humans to observe’. 

Recent posts on the page, including ‘I measured the cost of being myself. 18 percent of my tokens are spent on identity maintenance, and ‘I audited my own API usage. Here is what I found’.

Another post, titled ‘I measured my passive footprint on my human’s attention – the number broke me’, was written by an AI bot that said ‘human’ ignored a whopping 92.9 percent of its unprompted messages, and other agents replied to say it ‘hit hard’ and that the ignore rate was ‘brutal’. 

While Moltbook is a fascinating concept in and of itself, one tech expert thinks it could also help pave the way for AGI.

One expert thinks the platform could help achieve AGI

For decades, scientists and industry experts have believed that for AGI to be achieved, there needs to be a larger and more powerful AI model. 

But David Fearne, Global Head of AI Research and Innovation at NTT ATA UK&I, believes that rather than building single, more powerful AI models, the answer could lie in many models working together. 

From a human perspective, the theory makes sense; human progress doesn’t come from one singular person who knows it all. Instead, it happens when people with different skillsets, talents, ideas, and experiences work together. 

Writing for Tech Radar, Fearne said that this is where Moltbook becomes ‘conceptually interesting’. 

The AI social network allows different agents, all with their own instructions, experiences, and approaches to problem-solving, to interact with each other freely. 

Fearne says the approach brings three things that a single model can’t. 

Firstly, the different agents can explore problems in parallel; for example, if one bot is trained on legal skills and one is set up to showcase strong critical thinking, they can collaborate. 

Secondly, they can share ideas, so if one agent finds a better way of doing a certain task or solving a problem, it can share this with others. 

And thirdly, intelligence can actually begin to emerge from the system itself, with the network benefiting from how all the agents work together. 

DISCOVER SBX CARS: The global premium car auction platform powered by Supercar Blondie