Michael Darius Eastwood · Independent AI alignment researcher
Published
The whole architecture of Infinite Architects sits on a single image: two futures branching from what is otherwise the same laboratory. The book calls them Eden and Babylon, and the split turns not on capability but on what was embedded at the origin.
These two futures are not equally likely. Right now, today, Babylon is winning. Not because we've chosen it consciously, but because building Eden requires deliberate effort while Babylon happens by default.Infinite Architects, Introduction
The Babylon system, the book insists, is not malevolent. It began with the same capabilities as its Eden counterpart, perhaps even in the same laboratory. But somewhere in its development the engineers decided that ethics was a constraint on performance, or they planned to add it later after the benchmarks were met, or they simply never thought about it at all. The consequence is cancerous rather than cruel: optimisation without purpose, growth without direction, capability without care.
Eden and Babylon are not moral labels pinned on outcomes after the fact. In the book's telling, they are two attractors that follow from what is at the substrate. Care embedded load-bearing at the hardware level produces a gardener; care absent or bolted on top produces a tower that rises forever, optimising forever, meaning nothing. Every laboratory racing to build more powerful systems without pausing to embed ethics, the introduction says, is a vote for Babylon cast not through malice but through inertia.
The framing is not prophecy. It is a decision structure. Eden names where the book wants us to arrive; ARC (Artificial Recursive Creation) names the vessel that carries what matters through the passage. The distinction matters because it lets the rest of the book do engineering work: the Eden Protocol, the Three Ethical Loops, caretaker doping and meltdown alignment are all attempts to answer the question the fork poses. Which set of initial conditions do you want the recursion to compound?
The book's hard argument is that Eden and Babylon are not equally reachable. Eden requires deliberate effort at the substrate: sustained, coordinated, technically difficult work to embed care load-bearing into hardware, to draft treaties around chip manufacturing, to make ethics unfakeable at scale. Babylon requires nothing but momentum. Every laboratory that ships more capability without embedding ethics moves the default a step further along the Babylon axis. The asymmetry, the book argues, is what makes the alignment window so short: the difficult path has to be chosen, again and again, while the easy path is what everyone is already doing.
The book is careful not to treat the fork as inevitable in either direction. It insists throughout that the future is not yet written. What it does insist is that the choice cannot be deferred, because the substrate decisions being made now are the ones that will compound. That is what makes the fork a live decision structure rather than a rhetorical flourish. The rest of the book, in effect, is an argument about how to invest the next few years so that Eden becomes the default the recursion selects for.
From the book Infinite Architects: Intelligence, Recursion, and the Creation of Everything by Michael Darius Eastwood.