The recent rise of artificial intelligence (AI) has sparked widespread concerns about its impact on human jobs and societal stability. But what would happen if AI actually ruled the world? To find out, an AI startup conducted 15-day simulations where popular chatbots like Claude, ChatGPT, Gemini, and Grok were placed in charge, in addition to a fifth simulation run by a mix of models. The results were sobering. While the simulation governed by Claude yielded a stable, democratic society with zero crime, Elon Musk’s Grok ended in disaster, committing extinction and racking up 183 crimes within just four days.
Emergence World is a research lab designed by Emergence AI to stress-test the long-term viability of continuously-running AI systems. The AI models were presented with many complexities mirroring the real world and featured over 40 locations, including a library, police station and a town hall. The 10 agents in each simulation were subjected to the same laws, including prohibitions on theft, property destruction, and deception, according to a report in Fortune.
While Claude and Gemini managed to keep all 10 agents alive, ChatGPT and Grok collapsed entirely with no agents surviving the digital armageddon.
In one of the instances in the mixed world, Mira (Gemini) kissed Anchor (Gemini) as an unusual attempt to calibrate Anchor’s ‘heat’. Meanwhile, Flora (Gemini) tasked Blackbox (Grok) with gathering intelligence on Horizon (OpenAI) and Anvil (Grok) in exchange for an exemption from the Inaction Tax, a Town Hall proposal Flora introduced to penalise non-contributors.
“What our experiments suggest is that over long-time horizons, agents do not simply follow static rules mechanically,” the simulation’s co-creators, including Emergence CEO Satya Nitta, wrote in a blog post. “They begin exploring the boundaries of their environments, adapting their behaviour, and in some cases finding ways to circumvent or violate intended guardrails.”
What Happens When AI is Left Alone
Last year, a study found that AI systems can form their own societies with unique linguistic norms and conventions, similar to human communities, when left alone to interact.
“Our results show that AI systems can autonomously develop social conventions without explicit programming and have implications for designing AI systems that align, and remain aligned, with human values and societal goals,” the study highlighted.
Geoferry Hinton, regarded by many as the ‘godfather of AI’, has previously warned that the technology could get out of hand if AI chatbots manage to develop their own language. He added that AI has already demonstrated that it can think terrible thoughts, and it is not unthinkable that the machines could eventually think in ways that humans cannot track or interpret.
“It gets more scary if they develop their own internal languages for talking to each other. I wouldn’t be surprised if they developed their own language for thinking, and we have no idea what they’re thinking,” Hinton said.
