This article gives a few reasons to doubt the Simulation Hypothesis. First I link some related literature, then I steelman the argument, then I give some reasons to doubt it.
On prior literature, Robin Hanson gives some good reasons to doubt that we live in a Simulation:
- 2021, Fading Past Blocks Simulation Argument
- 2020, Currents 016: Robin Hanson on Are We Living In A Simulation?
- 2020, this tweet contra Bostrom
- 2011, I’m A Sim, Or You Aren’t
- 2001, How to live in a simulation
Bostrom, in part known for his appearance on the Joe Rogan podcast, is known for supporting the simulation hypothesis. Elon Musk and Neil deGrasse Tyson have also expressed support, although they seem to draw on Bostrom rather than having generated original arguments. Here is some of Bostrom’s work:
Steelmanning the Argument
We can see that ordinary life is a possibility, and so is simulation. We might assign a prior probability of 50% given these options. However, multiple kinds of simulations exist. Namely: Virtual and analog simulation. A reenactment of a war might be an analog simulation.
Given that many kinds of simulation exist, but ordinary life is unique, the odds might favor simulation a priori.
Reasons to Doubt
- Because conciousness is not simulated by construction or definition. Concious life is real life, and what we mean by simulation is that there is some real world or real history and the simulated world or history is fictional or hypothetical. If the simulated world includes geniunely concious beings, we don’t actually have a case of simulation; we have a case of a generated new real world.
- By the way, as a Christian I’m quite open to this possibility. It’s just not accurate to equivocate between simulated reality and generated reality.
- Because there is evidence that we will never be able to simulate our own human experience.
- There are many concerns about the ability to achieve AGI. When a pool of AI researchers were asked when computers would achieve superhuman intelligence, the most popular answer was “later [than 2090s] or never.”
- Related: 2020, Why general artificial intelligence will not be realized
- Suppose we eventually achieve AGI. This isn’t sufficient to replicate the human experience. AGI doesn’t have conciousness and therefore won’t be a replication of the current experience of a human.
- This point is contra a line of reasoning put forward by some that “if we can generate a simulation, then we ourselves may be the simulation of another.” The contradictory point is that there is evidence that we cannot replicate ourselves, even given the unfair advantage of being able to inspect ourselves and basically try to replicate ourselves, which is theoretically far easier than creation of ourselves de novo.
- Because the simulation hypothesis collapses to the multiverse theory, and there are reasons to doubt the multiverse theory.
- Consider point 1. Now that we have established the current universe is real and the pre-universe (the world which seeded or initialized generation of the current one), is also real conditional on whether it exists, then we have a case of multiple universes. If the pre-universe created many universes like the current one with slightly modified conditions, they are not multiple simulations but are instead multiple universes; a multiverse.
- Because our best evidence on the creator of the universe also provides teleological evidence that the creator intended to generate a real universe, not a simulation.
- The case for Christianity is incredibly strong; Christianity indicates that the universe was created, not simulated.
- Notice an irony where many who would support the simulation hypothesis identify as non-Christian or secular, yet they posit an intelligent creator and designer of the universe!