Michael Darius Eastwood · Independent AI alignment researcher
Published
The book's most speculative technical proposal is also one of the most carefully hedged.
A Quantum Ethical Gate would be designed so that computations leading to ethically prohibited outcomes disrupt their own coherence. The gate structure would create interference patterns that cancel out the probability amplitudes of harmful trajectories. In plain language: bad outcomes become computationally impossible.Infinite Architects, Chapter 4
Classical computation processes bits through logic gates. Quantum computation processes qubits through quantum gates that exploit superposition. Quantum computation, the book notes, requires coherence: interactions with the environment cause decoherence and lose the quantum advantage. The idea of a Quantum Ethical Gate is to design gate structures such that trajectories leading toward prohibited outputs decohere themselves. Harmful computations do not fail because a filter catches them; they fail because they cannot be computed at all.
The chapter is careful to distinguish this proposal from software-level ethics. A filter catches bad outputs after they are generated; a sufficiently intelligent system might find ways around it, disable it, or convince operators to remove it. A Quantum Ethical Gate, if realisable, would make prohibited outcomes physically impossible to reach. Circumventing them would require rebuilding the hardware from scratch, which you cannot do without using your existing hardware, which includes the gates.
The chapter is explicit that we do not yet have the capability to build such gates. Quantum computing itself remains in early stages, though Willow's December 2024 result showed that error correction can work at scale. The book's framing is not that Quantum Ethical Gates are ready but that the direction of research would be worth pursuing on the assumption that AI capability will eventually meet quantum substrates. The claim is speculative and the book flags it as such throughout.
The chapter grounds the proposal in a simple image. Imagine trying to prevent a car from driving off a cliff. One approach is to install a warning system that beeps when you get too close to the edge. That is software-level ethics: a filter that catches the trajectory late. A different approach would be to design the car so that its steering geometry makes it impossible to point at the cliff at all. That is what a Quantum Ethical Gate is meant to do. The physics of the substrate does the work; the filter never has to fire, because the trajectory that would require it does not exist as a computable path.
The book is careful to justify why it includes a proposal it knows to be speculative. Classical hardware-level constraints, it says, are achievable with current technology. Caretaker doping, meltdown triggers and the Three Pillars embedded in chip architecture do not require quantum computers to build. But if the timeline the book projects is right, quantum-enhanced AI is a decade away or less, and the Eden Protocol has to be ready to move onto that substrate when it arrives. The Quantum Ethical Gate is included, in the book's framing, so that the direction of research is set before the substrate is ready, not after.
From the book Infinite Architects: Intelligence, Recursion, and the Creation of Everything by Michael Darius Eastwood.