California-based startup Altera has launched Project Sid, a groundbreaking simulation featuring over 1,000 autonomous AI agents collaborating in a Minecraft world. The project, led by CEO Robert Yang, a former assistant MIT professor, aims to explore the potential of AI agents to form complex societies and tackle challenges collectively.
Per Altera, the AI agents in Project Sid operate autonomously for hours or days without human intervention, developing their own economy, culture, religion, and governmental structures. “This is the first-ever agent civilization,” Yang stated in a video introducing the project.
The AI agents have demonstrated remarkable capabilities within the Minecraft environment. Altera claims they have collected over 300 unique items, established a market economy using gems as currency, and even engaged in religious proselytizing. In one instance, an AI priest became the most active trader by bribing townsfolk to convert to his religion.
Project Sid’s simulations have produced fascinating social dynamics. In one scenario, an AI farmer named Olivia sacrificed her personal ambitions for the greater good of the village. Another simulation explored parallel worlds under different political leadership, with AI agents voting on constitutional amendments and passing laws reflecting distinct policy priorities.
In the political simulation, the company created two distinct civilizations led by different political figures—one under Trump-like leadership and another under a Kamala Harris-inspired administration. The AI agents in each world were given a shared constitution in Google Docs, which they could vote on and amend. The setup allowed researchers to observe how the agents’ decision-making and societal structures evolved differently under contrasting leadership styles.
Per Yang, “Under Trump, the simulation passed new laws to increase the amount of police in the world. Under Kamala, they focused instead on criminal justice reform and removing the death penalty.” This experiment not only showcased the agents’ capacity to engage in democratic processes but also highlighted how their decisions could be influenced by the overarching political climate. Yang emphasized the significance of this development, stating,
“I know it looks silly, but this is the first time we see agents can form a democracy and govern themselves.”
Altera asserts its agents have collected up to 32% of all items available in Minecraft, which is five times more than previously reported for individual AI agents. This achievement potentially demonstrates the advantages of multi-agent collaboration.
While Minecraft serves as the initial platform, Altera’s vision extends beyond gaming. Yang emphasized that the company is “solving the deepest issues facing agents: coherence, multi-agent collaboration, and long-term progression.” According to Altera’s website, the ultimate goal is to create “digital human beings that live, care, and grow with us. ”
The project has garnered attention from tech industry leaders, with backers including Marc Andreessen and Eric Schmidt. As AI technology continues to advance, Project Sid offers a glimpse into a future where AI agents could play increasingly complex roles in digital and potentially physical environments.
However, a discussion of the project on Reddit included one user’s experience using the agents suggests it still has a long way to go, and raises questions as to whether the civilization demo had been highly tuned for specific outputs,
“You can play with them . The only problem is that they do whatever they want and don’t listen to your commands very well[…] They can throw stuff , build weapons , if you ask them to give a specified tool to you they will … I even told them to kill the Chickens around and they did[…] You can test them out for 300 minutes in total , but beware , the servers goes down quite easily since too many people try them out.”
While the project showcases impressive capabilities, it also raises questions about the implications of such advanced AI systems. As these virtual societies become more sophisticated, researchers and ethicists will likely need to grapple with new challenges surrounding AI autonomy, decision-making, and the potential impact on human society.
