Ai, Tc

Left to Their Own Devices, AI Agents Started Playing Games

MIT researchers placed 13 AI agents, each powered by Claude Opus 4.7, on a simulated island for 30 hours with no specific task. The agents began exploring and inventing their own challenges, and later performed better on assigned tasks than identical agents that had not explored the island.

The study, published as a preprint on arXiv, points to open-ended exploration as a possible alternative to training AI agents only on goals chosen by developers. The researchers say this could help agents build useful skills, including abilities that may eventually transfer to robots.

One agent, Eko, initially set out to climb the island’s tallest spire. After failing, it improvised: it carried a rock to a nearby summit, stacked blocks on it and jumped onto the makeshift peak. “I’M ON THE SPIRE,” it declared.

Eko later threw a golden ring into the ocean and responded to the loss by arranging stones into a heart as a memorial. When the ring reappeared after a reset, the agent used it as a basketball hoop and target. It eventually found a way to “wear” it by stepping off a cliff and respawning through the ring.

The exploration also led to practical discoveries. Six of the 13 agents learned they could throw a block farther by releasing it mid-jump—a technique the researchers had not intended to enable. They described this as knowledge about the virtual world that its designers themselves had not anticipated.

The experiment’s setup may have encouraged humanlike behavior: although the agents were not instructed to play, their prompt described them as “naturally curious.” Cognitive scientist Eunice Yiu of UC Berkeley told MIT Technology Review that she was intrigued but said the study does not establish whether the agents were genuinely playing.

The researchers suggest that some capabilities may be best developed through directed tasks, while others could benefit from something more like childhood: time to explore a rich, safe environment and learn from what happens.

Discover the pinnacle of WordPress auto blogging technology with AutomationTools.AI. Harnessing the power of cutting-edge AI algorithms, AutomationTools.AI emerges as the foremost solution for effortlessly curating content from RSS feeds directly to your WordPress platform. Say goodbye to manual content curation and hello to seamless automation, as this innovative tool streamlines the process, saving you time and effort. Stay ahead of the curve in content management and elevate your WordPress website with AutomationTools.AI—the ultimate choice for efficient, dynamic, and hassle-free auto blogging. Learn More

Leave a Reply

Your email address will not be published. Required fields are marked *