There was a mathematical/probabilistic argument proposed in 2003 [1] that, if I were to paraphrase, goes like this:
* It's a virtual certainty that hardware and software improvements will lead to us being able to simulate beings like ourselves one day, with the simulations being so good that the simulated beings will not know they are simulated.
* Once that happens, we will create a large number of such simulations and simulated beings.
* At that point, given all beings who do not think they are simulated (or don't know for sure, like us), the probability that a being is simulated is pretty close to 1 and the probability that a being is real is close to zero. [Frequentist argument based on number of beings of each type]
* Finally, there's no reason to believe this has not already happened.
* It's a virtual certainty that hardware and software improvements will lead to us being able to simulate beings like ourselves one day, with the simulations being so good that the simulated beings will not know they are simulated.
* Once that happens, we will create a large number of such simulations and simulated beings.
* At that point, given all beings who do not think they are simulated (or don't know for sure, like us), the probability that a being is simulated is pretty close to 1 and the probability that a being is real is close to zero. [Frequentist argument based on number of beings of each type]
* Finally, there's no reason to believe this has not already happened.
[1]: http://www.simulation-argument.com/simulation.html
EDIT:formatting.