Automation, Not Autonomy: Rethinking Agent-Driven Autonomous World
Preliminary Statement: I’m not an anti-Agent advocate.
Okay, let’s get to the point.
Right now, 80% of Autonomous Worlds (AW) are actually Agent-driven worlds. Perhaps it’s time we redefine the “A” in AW — not as “Autonomous” but as “Automated.” Welcome to the Autonomous vehicles. (Joking, but only slightly.)
As an AW builder, I fully acknowledge the role of AI in shaping the future of AW — AI isn’t just useful; it’s essential. But here’s the real question: Are today’s AW truly Web3, or are they just Web2 simulations with Agent at the center?
If an Agent World lacks any meaningful blockchain integration, does it really qualify as an Autonomous World? Most of these AI-driven systems rely on centralized infrastructures for processing world events. Where does “autonomy” fit into that picture?
It’s important to define what we mean by “autonomy” in AW:
• Is it about governance?
• Is it about user-generated content (UGC), where players shape the world?
• Is it about a self-sustaining system?
Previously, AW achieved self-sustainability through on-chain mechanisms. Now, the paradigm has shifted to AI-driven automation. But what’s the purpose of an AW if the only thing we do is watch Agents operate in loops?
AI has incredible utility in AW — especially in on-chain transactions, cross-chain interoperability, and NPC behaviors. But we need to ask ourselves: What are we actually building? As a friend once told me, many of today’s AW feel just like GameFi 1.0 — repetitive and boring.
Get Raphael Joe’s stories in your inbox
Join Medium for free to get updates from this writer.
Seriously, who wants to watch Agent running around aimlessly? Click-to-Earn at least required some player interaction, even if the rewards were meaningless. But watching data evolve? Is that really engaging?
Many of these Agent-driven worlds aren’t actually trying to build AW — so they don’t care about critical elements like digital physics or emergent player dynamics.
As AI becomes the dominant driver of AW logic, let’s consider what Claude (Anthropic’s LLM) has to say:
“The biggest flaw of Agent Worlds is that they attempt to replicate real-world complexity using mechanical logic, when true intelligence and autonomy require emergent, self-organizing behaviors.”
Now, look at Lootverse. The original Loot project might not be what it once was, but the ecosystem has continued to evolve organically. Artists, designers, engineers, enthusiasts, and players — real humans — are driving its growth.
I still believe Agents have revolutionary applications in AW beyond just NPC interactions. But their role should be transformative — not as the core operating system of the world. Just like NFTs in gaming aren’t meant to be mere collectibles — they should be something far more disruptive.
I respect the Agent-centric AW, but I believe the “A” in AW should stand for “Autonomous,” not just “Automated.”
However, we believe AW should be truly autonomous, not just automated. Instead of relying solely on Agents to dictate world dynamics, we’re designing digital frameworks, emerging interactive experiences, and digital physics — ensuring that AW remain decentralized, engaging, and meaningful.
Because an Autonomous World should be more than just Agent running in circles. It should be a living system where AI, users, and blockchain converge to create something far greater than the sum of its parts.
