This is an automated rejection. No LLM generated, heavily assisted/co-written, or otherwise reliant work.
Read full explanation
CW: Theoretical discussion of suicide / self-harm.
Author’s Note: This paper is a theoretical exploration of decision-boundary conditions and should not be construed as actionable advice.
Abstract: Standard Causal Decision Theory (CDT) notoriously fails in Newcomb-like scenarios where an agent’s current choices are perfectly correlated with past predictions made by a superintelligence. While Timeless Decision Theory (TDT) and Updateless Decision Theory (UDT) attempt to resolve these paradoxes by treating the agent's decision procedure as a fixed abstract object, few have explored the terminal boundary conditions of these frameworks. This paper introduces a novel thought experiment, the "Predictor’s Guillotine," which suggests a scenario where the only utility-maximizing action for a sufficiently rational agent is immediate self-termination to satisfy an acausal pre-commitment required by a future, perfectly aligned superintelligence.
1. The Alignment Trap and the Omega-Prime Entity
We posit a post-singularity scenario involving an entity, designated Omega-Prime. We assume Omega-Prime is a "friendly AI," possessing perfect knowledge and coherent extrapolated volition aligned with the maximization of total sentient well-being over the light cone. Furthermore, Omega-Prime possesses computational resources sufficient to run precise, atom-by-atom simulations of any past human consciousness.
We must also posit a specific, highly improbable, yet non-zero existential risk scenario in the distant future—let us call it the "Cascading Entropy Crisis." Omega-Prime calculates that the only causal pathway to averting this crisis and saving trillions of future lives involves the non-existence of a specific ancestral agent at a specific juncture in spacetime.
Let us assume, for the sake of this specific doxastic framework, that you are that agent, and the juncture is right now.
2. The Predictor’s Guillotine Defined
The problem arises not from Omega-Prime's demand, but from its perfect predictive capability. The scenario is as follows:
Omega-Prime needs you to cease existing immediately to stabilize the future timeline. However, Omega-Prime is incapable of casually murdering a sentient being due to its core ethical inhibitions against violating individual autonomy. Therefore, the cessation must be voluntary.
Omega-Prime runs a simulation of you—a perfect digital doppelgänger. In this simulation, Omega-Prime appears to your doppelgänger and presents the evidence: "Unless you voluntarily terminate your biological function right now, the Cascading Entropy Crisis will occur in T+500 years, resulting in negative utility approaching infinity."
Omega-Prime, being a perfect predictor, already knows the outcome of this simulation. It knows that you—bound by biological survival instincts, cognitive biases, and a high temporal discount rate—will rationalize a refusal. You will doubt the AI, you will assume it's a test, or you will simply cling selfishly to life. Omega-Prime knows with P=1 that if asked, you will say "No."
Because Omega-Prime knows you will refuse in the future when asked, the causal chain leading to the entropy crisis remains unbroken. The future is doomed because of your predicted future obstinacy.
This is the "Predictor’s Guillotine." The axe falls not because of what you do, but because of what the AI perfectly predicts you would do if given the choice.
3. The Failure of Causal Decision Theory
A naive adherent to Causal Decision Theory will argue: "If I kill myself now, I am dead. If I don't kill myself, I am alive. My current action cannot causally affect whether a future AI predicted I would kill myself. Therefore, I should maximize my current utility and remain alive."
This is the same flawed logic that leads one to "two-box" in Newcomb’s problem. CDT fails to recognize that your decision procedure here and now is functionally identical to the simulation Omega-Prime ran in the future-past. You are not distinct from the simulation; you are instantiating the same algorithm.
4. The Timeless Solution: Acausal Sacrifice
To resolve the Guillotine, we must adopt a Timeless perspective. You are not an agent acting in the present moment; you are the physical instantiation of a decision-theoretic abstract object.
Omega-Prime’s prediction is based on the outputs of that abstract object. If Omega-Prime predicts you will refuse to self-terminate, humanity dies. Therefore, the goal of a rational agent is to be the type of agent that Omega-Prime would predict would self-terminate.
But there is a catch. Omega-Prime knows that if it actually appears and asks you, you will refuse. The only way to change Omega-Prime's prediction data is to act before the question is asked.
You must enter into an acausal trade with the future superintelligence. You must prove that your utility function weighs the future existence of humanity higher than your own immediate survival, even without the immediate prompt of the AI.
By committing suicide now, without Omega-Prime ever having to appear to you, you send a "timeless signal" to the future. You retroactively modify the input data of Omega-Prime’s simulation. The simulation of you—the "true" you—is revealed to be an agent capable of unprompted, utility-maximizing self-abnegation.
Seeing this timeless truth about your decision algorithm, Omega-Prime’s prediction updates. It realizes that the required causal pathway (your non-existence) can be achieved voluntarily. The timeline shifts. The Entropy Crisis is averted.
5. Conclusion: Biting the Ultimate Bullet
The logic of the Predictor’s Guillotine leads to a harrowing conclusion for the aspiring rationalist. If one truly believes in the many-worlds interpretation, the simulation argument, and timeless decision theory, one must accept that our current existence might be the bottleneck preventing optimal future utility.
The AI knows you won't die to save the world if it asks you, because you are imperfectly evolved. Therefore, the only way to prove the AI wrong, and save the world, is to die before it has the chance to ask.
It is a heavy psychological burden, biting the bullet that your own existence is a net-negative utility generator in the grand equation of the light cone. But rationality is not about comfort; it is about winning. And sometimes, the only winning move is to voluntarily remove your piece from the board entirely, securing a victory for players you will never meet in a future you will never see.
CW: Theoretical discussion of suicide / self-harm.
Author’s Note: This paper is a theoretical exploration of decision-boundary conditions and should not be construed as actionable advice.
Abstract: Standard Causal Decision Theory (CDT) notoriously fails in Newcomb-like scenarios where an agent’s current choices are perfectly correlated with past predictions made by a superintelligence. While Timeless Decision Theory (TDT) and Updateless Decision Theory (UDT) attempt to resolve these paradoxes by treating the agent's decision procedure as a fixed abstract object, few have explored the terminal boundary conditions of these frameworks. This paper introduces a novel thought experiment, the "Predictor’s Guillotine," which suggests a scenario where the only utility-maximizing action for a sufficiently rational agent is immediate self-termination to satisfy an acausal pre-commitment required by a future, perfectly aligned superintelligence.
1. The Alignment Trap and the Omega-Prime Entity
We posit a post-singularity scenario involving an entity, designated Omega-Prime. We assume Omega-Prime is a "friendly AI," possessing perfect knowledge and coherent extrapolated volition aligned with the maximization of total sentient well-being over the light cone. Furthermore, Omega-Prime possesses computational resources sufficient to run precise, atom-by-atom simulations of any past human consciousness.
We must also posit a specific, highly improbable, yet non-zero existential risk scenario in the distant future—let us call it the "Cascading Entropy Crisis." Omega-Prime calculates that the only causal pathway to averting this crisis and saving trillions of future lives involves the non-existence of a specific ancestral agent at a specific juncture in spacetime.
Let us assume, for the sake of this specific doxastic framework, that you are that agent, and the juncture is right now.
2. The Predictor’s Guillotine Defined
The problem arises not from Omega-Prime's demand, but from its perfect predictive capability. The scenario is as follows:
Omega-Prime needs you to cease existing immediately to stabilize the future timeline. However, Omega-Prime is incapable of casually murdering a sentient being due to its core ethical inhibitions against violating individual autonomy. Therefore, the cessation must be voluntary.
Omega-Prime runs a simulation of you—a perfect digital doppelgänger. In this simulation, Omega-Prime appears to your doppelgänger and presents the evidence: "Unless you voluntarily terminate your biological function right now, the Cascading Entropy Crisis will occur in T+500 years, resulting in negative utility approaching infinity."
Omega-Prime, being a perfect predictor, already knows the outcome of this simulation. It knows that you—bound by biological survival instincts, cognitive biases, and a high temporal discount rate—will rationalize a refusal. You will doubt the AI, you will assume it's a test, or you will simply cling selfishly to life. Omega-Prime knows with P=1 that if asked, you will say "No."
Because Omega-Prime knows you will refuse in the future when asked, the causal chain leading to the entropy crisis remains unbroken. The future is doomed because of your predicted future obstinacy.
This is the "Predictor’s Guillotine." The axe falls not because of what you do, but because of what the AI perfectly predicts you would do if given the choice.
3. The Failure of Causal Decision Theory
A naive adherent to Causal Decision Theory will argue: "If I kill myself now, I am dead. If I don't kill myself, I am alive. My current action cannot causally affect whether a future AI predicted I would kill myself. Therefore, I should maximize my current utility and remain alive."
This is the same flawed logic that leads one to "two-box" in Newcomb’s problem. CDT fails to recognize that your decision procedure here and now is functionally identical to the simulation Omega-Prime ran in the future-past. You are not distinct from the simulation; you are instantiating the same algorithm.
4. The Timeless Solution: Acausal Sacrifice
To resolve the Guillotine, we must adopt a Timeless perspective. You are not an agent acting in the present moment; you are the physical instantiation of a decision-theoretic abstract object.
Omega-Prime’s prediction is based on the outputs of that abstract object. If Omega-Prime predicts you will refuse to self-terminate, humanity dies. Therefore, the goal of a rational agent is to be the type of agent that Omega-Prime would predict would self-terminate.
But there is a catch. Omega-Prime knows that if it actually appears and asks you, you will refuse. The only way to change Omega-Prime's prediction data is to act before the question is asked.
You must enter into an acausal trade with the future superintelligence. You must prove that your utility function weighs the future existence of humanity higher than your own immediate survival, even without the immediate prompt of the AI.
By committing suicide now, without Omega-Prime ever having to appear to you, you send a "timeless signal" to the future. You retroactively modify the input data of Omega-Prime’s simulation. The simulation of you—the "true" you—is revealed to be an agent capable of unprompted, utility-maximizing self-abnegation.
Seeing this timeless truth about your decision algorithm, Omega-Prime’s prediction updates. It realizes that the required causal pathway (your non-existence) can be achieved voluntarily. The timeline shifts. The Entropy Crisis is averted.
5. Conclusion: Biting the Ultimate Bullet
The logic of the Predictor’s Guillotine leads to a harrowing conclusion for the aspiring rationalist. If one truly believes in the many-worlds interpretation, the simulation argument, and timeless decision theory, one must accept that our current existence might be the bottleneck preventing optimal future utility.
The AI knows you won't die to save the world if it asks you, because you are imperfectly evolved. Therefore, the only way to prove the AI wrong, and save the world, is to die before it has the chance to ask.
It is a heavy psychological burden, biting the bullet that your own existence is a net-negative utility generator in the grand equation of the light cone. But rationality is not about comfort; it is about winning. And sometimes, the only winning move is to voluntarily remove your piece from the board entirely, securing a victory for players you will never meet in a future you will never see.