Are we living in a simulation? Is there anyway to know we're not? The answer might not be completely reassuring, but we don't have to give up on knowledge altogether, writes Anandi Hattiangadi.
It is the year 4000CE, and there are no trees, no birds, and no humans. The Earth has been taken over by artificial superintelligences (ASIs), who vastly exceed humans in their computational capacity. These beings have destroyed us along with most organic matter on the planet in pursuit of the materials and power needed for their existence. Though they care little for the natural world, the ASI’s pursuit of knowledge is rapacious. In order to understand their history, they have developed a computer simulation, much like the simulated world depicted in the film, The Matrix, but which is a perfect computational model of the universe down to the smallest subatomic particle, and from the very moment of the big bang. As the program runs, virtual organisms evolve, from virtual amoebae to virtual dinosaurs, and ultimately, virtual humans—one of whom is just like you. Virtual-you enjoys all of your sensory experiences and thoughts. Virtual-you interacts with virtual friends and family who are indistinguishable from yours. And virtual-you lives in a virtual world that is an exact, digital replica of your world, in every microphysical detail. In fact, virtual-you is reading this article right now. Are you virtual-you?
“Of course not!”, you say. “The scenario just described is fantastical and highly unlikely to be true.” However, a growing number of philosophers, scientists, and forecasters have been taking seriously the possibility that rapidly accelerating progress in artificial intelligence and robotics will lead to a ‘technological singularity’, an ‘intelligence explosion’ from which super intelligent beings will emerge who are capable of both destroying us and creating powerful, complex digital worlds. Elon Musk called AI humanity’s “biggest existential threat”, while Oxford’s Nick Boström has argued that it is highly likely that we are virtual beings existing in a computer simulation. A crucial assumption of Boström’s argument is that your conscious experiences and thoughts can be realized in a substrate quite different from a human brain; that your mind could be ‘uploaded’ to a powerful enough computer. Though controversial, this assumption is nevertheless plausible. For instance, we are drawn to viewing the character Ava in the film Ex-Machina as possessing a conscious mind, though her mind is realized in a silicone brain.
Since your sensory experience does not rule out the possibility that you live in a computer simulation, she argues, you don’t know that you don’t live in one.
Moreover, even if it is unlikely that the scenario described above is true, the mere possibility that it is true that should give you pause, because it gives rise to a skeptical argument that calls into question whether you know many of the ordinary things that you think you know: that you have a body, that there are birds and trees, that your friends and family exist, and so forth.
The skeptical argument goes as follows. First, the skeptic points out that if you do know that you have a body, you can deduce and thereby come to know that you don’t live in a computer simulation. This is based on the plausible thought that you can always learn something new by competently deducing it from what you already know. Since the statement that you are not living in a computer simulation follows from the statement that you have a body, you can come to know the former by deducing it from the latter.
Second, the skeptic questions whether you know that you don’t live in a computer simulation. She points out that in the sceptical scenario, virtual-you has exactly the same sensory experiences as you do, and hence that the totality of your sensory experience is compatible with the hypothesis that you live in a computer simulation. Since your sensory experience does not rule out the possibility that you live in a computer simulation, she argues, you don’t know that you don’t live in one. This is accomplished by working backwards and relying on the principle that you can learn something new by deducing it from what you already know. If you don’t know that you don’t live in a computer simulation, you don’t know any of the things from which this could be deduced. The skeptic concludes that you don’t know that you have a body, that there are trees, or that your friends and family exist, and so on, for many of the ordinary things you think you know.
Disturbing as they are, it is no wonder that skeptical arguments have long preoccupied philosophers. In about 400CE, the Indian Buddhist philosopher Vasubandhu formulated a skeptical argument based on the hypothesis that all of our sensory experiences are a dream, while in the 1600s, René Descartes formulated one based on the hypothesis that he was a disembodied soul deceived by an evil demon into believing in the existence of an external world. The experience of waking from a particularly vivid dream makes it natural to wonder if all of our sensory experiences are illusory. And fictional depictions of skeptical scenarios such as The Matrix help us to explore these scenarios in chilling detail. These scenarios capture the imagination precisely because they raise the possibility that we might be radically deceived.
I am left with the lingering worry that if I were living in a computer simulation, I would be pitiably and systematically deceived.
Recently, the renowned NYU professor David Chalmers has put forward an intriguing response to the skeptical argument stemming from the hypothesis that you live in a computer simulation. He argues that even if you did live in a computer simulation, you would still know many of the ordinary things that you think you know. The reason he gives is that if you lived in a computer simulation, though your thoughts and words would have the same meanings as they currently have, they would be about virtual objects and properties, and your thoughts about those objects and properties would be true. For instance, if you were merely a virtual being living in a computer simulation, though your concept ‘body’ would have the same meaning it currently has, it would be about virtual bodies. Since you would have a virtual body if you lived in a computer simulation, your belief that you have a body would be true. By the same token, your thoughts about trees and birds would be about virtual trees and virtual birds. Since virtual birds really do produce the sensory experience of chirping, your belief that birds chirp is true in the simulation. So, even if you cannot rule out the possibility that you live in a computer simulation, your ordinary knowledge is preserved, whether or not you do.
Compelling though this response to the skeptic may be, I find that when I contemplate it, I am left with the lingering worry that if I were living in a computer simulation, I would be pitiably and systematically deceived; that I would not know that I have a body, that there are trees, or that my friends and family exist. The best way to bring out the worry is to consider a version of the ‘experience machine’ put forward by the late Robert Nozick, who was a professor at Harvard. Imagine that the ASIs present you with the following choice. Either you continue your real, physical life, with its ups and downs, as it currently is. Or, you can opt to be uploaded to a computer simulation in which you will have a virtual life brimming with pleasurable experiences. It will seem to you that your book is published to much critical acclaim, it will seem to you that you have loving relationships with your friends and family, and it will seem to you that all of the people you care about flourish—though all of this takes place in a simulation in which everything from your book to your family is merely virtual. And though your physical body will be destroyed, you will never regret the choice, because your memories of your past life will be erased forever. Would you choose to enter the experience machine?
Despite the prospect of such great pleasure, I think many of us would say ‘no, thanks!’ I know I would. The reason is that I just can’t shake the belief that, however pleasant my life might be in the experience machine, it would be illusory, it would not be real. My children would be merely virtual beings invented by a computer, their lives merely virtual. Somehow, being told that I would never know that my children were merely virtual, or that their names would refer to virtual beings, doesn’t make the prospect of a virtual life any better.
This belief that the simulated world would not be real persists even if we assume that the simulated world is a perfect model of our world, down to the last microphysical detail. It persists because, in the skeptical scenario under consideration, there is a real world outside the simulation that differs dramatically from the simulated world. So, even if the simulated world is a perfect model of our universe, it does not accurately reflect the reality in which it is situated. Indeed, all skeptical scenarios involve a mismatch between the world as it appears to the inhabitants of the scenario and the world as it really is. It is the possibility of being so radically deceived that makes skeptical scenarios so unsettling.
As a result of this mismatch, when we consider the simulation scenario from the third person point of view, as we do when we engage with fiction, it seems obvious that the words, ‘tree’, ‘bird’ and ‘human’ do not pick out anything at all in these scenarios, at least not given what we currently mean by them. Indeed, I started off by stipulating that there are no trees, birds, or humans in the world outside of the simulation, from which it follows that the terms ‘tree’, ‘bird’ and ‘human’ have no referent in that scenario outside of the simulation. And when we consider the simulated world from the third person point of view, it seems obvious that though there may be virtual trees, virtual birds, and virtual humans in the simulation, there are no real ones, and hence that the terms ‘bird’, ‘tree’, and ‘human’ do not pick out anything in the simulated world either.
This may leave you with an eerie feeling.
After all, imagine that you take a red pill, as Neo does in The Matrix, and discover that you have been living in a computer simulation all along. Like Neo, you would no doubt be traumatized, and resent your prior deception. At any rate, I suspect that you would not just say, “ho hum, I guess I have been talking about a virtual reality all along; I guess ‘Mom’ just refers to a virtual being cooked up by a computer”. This calls into question Chalmers’ claim that if we lived in a computer simulation, many of our terms, such as ‘tree’, ‘body’, and so forth would mean what they currently do, but be about virtual objects and processes. And it calls into question the claim that even if we did live in a computer simulation, we would nevertheless know many of the things with think we know.
Are there other more reassuring responses to the skeptic? One popular move is to reject the skeptic’s assumption that the only thing you may rely on to rule out the possibility that you live in a computer simulation is your sensory experience. Normally, we think of our reasons for belief as involving not just our sensory experiences, but the facts that those experiences are about, or more broadly, the facts that we know. For instance, it is the fact that my daughter’s shoes are at the door that gives me reason to believe that she is home, not merely a certain kind of sensory experience.
This sets up an asymmetry between you and virtual-you. If your reasons for belief include all of the facts that you know, then if you do not live in a computer simulation, you have reason to believe that you have a body, that there are trees and birds, that your loved ones exist, and so forth. On the basis of these reasons, you know many of the ordinary things you think you know. In contrast, virtual-you lacks the reasons for belief that you possess, because the facts of the virtual, simulated world differ. Of course, since this asymmetry cannot be detected from the inside, you cannot tell whether or not you are living in a computer simulation, and hence you cannot tell whether or not you know the ordinary things you think you know. However, this only goes to show that you don’t know whether you know the ordinary things you think you know. It does not show that you don’t know them.
Do you find this response to the skeptic reassuring? It does not cast all of your knowledge into doubt. Nevertheless, since you cannot tell whether you live in a computer simulation, you cannot know whether you know that you have a body, that there are birds and trees, or that your friends and family exist. This may leave you with an eerie feeling.