Are We AIs?

Tom Easton

If Artificial Intelligence (AI) software is already close to being able to learn to be us by watching us, then we must be AIs.

This outrageous statement follows from physicist Frank Tipler’s 1994 proposal in The Physics of Immortality (Doubleday) and Nick Bostrom’s 2003 argument[1] that we live in a vast computer simulation. In essence, Bostrom says, we face three possible futures: we go extinct, our descendants don’t run many (if any) “ancestor” simulations, or we are living in a simulation. Such a simulation would be created by our descendants, millions or billions of years hence, in an effort to recreate the past. Such simulations might even be done many times, with variations in conditions, rather in the way “alternate history” writers construct stories based on the South winning the Civil War or Germany winning World War II.  There could also be a certain amount of evil laughter and “Let’s see how they handle this!

Bostrom concluded that because we are ignorant of the future, we are obliged to rate the three possibilities as roughly equal in probability. Elon Musk of Tesla and SpaceX fame was once asked what he thought of the idea that we are living in a computer simulation.  He replied that he thought it inevitable that reality-like simulations would someday be done and that because they would be done many times, there was only one chance in billions that we are living in “base reality.”[2] Of course, some people think both Bostrom and Musk are full of beans,[3] or that simulated minds (as well as AIs and minds downloaded into the computer) can’t possibly be conscious or “alive.” Since we are demonstrably both conscious and alive, we can’t possibly be living in a simulation.

This is the sort of discussion that makes lay people laugh at “pointy-headed intellectuals.” But academic sorts can not only take the idea seriously but get quite excited about it. Indeed, on April 5, 2016, at the 17th annual Isaac Asimov Debate held at the American Museum of Natural History in New York City, astronomer Neil deGrasse Tyson aimed four scientists and a philosopher at the question. Tyson himself rated the probability that we live in a simulation fairly high. The others just could not agree.[4]

If we do live in a simulation, then what does it mean? Is there any way to tell, the way you can tell a digital image from a view out a window by zooming in until you see the pixels[5]? Can we figure out how to break out of the simulation? Can we learn how to hack it (to win a lottery, a Nobel Prize, an election, or a war, among other things)? Probably no, no, and no, since we’re part of the simulation. It’s rather like asking whether your thigh bone can escape from your body. On the other hand, a cancer can escape—though only from one body to another, as shown by the Tasmanian devil’s contagious cancers.[6] I don’t think we’d be satisfied if we managed to escape the simulation, only to find that we had jumped to a different simulation. On the third hand, that would provide a nifty explanation for the many-worlds interpretation of quantum mechanics,[7] as well as the parallel-worlds trope in science fiction.

Does the answer even matter? After all, everything feels quite real. If it’s all a simulation, it’s a very good simulation. If it isn’t real, we can still pretend it is and get on with life.

Does it even matter whether we really had a “real” life in the distant past? If we did, then our present situation is an ancestor simulation and we live in a genuine, religion-free after-life[8] (unless you want to call the sysop or the programmer god, and then consider prayers as attempts at hacking). If we didn’t, then we have been constructed de novo within the simulation and are therefore AIs.  This also answers the question of whether AIs can truly be conscious or alive: we are both—we insist on it!—and if we are AIs, then…

Now let’s get confusing…

If AIs learn to be us by watching, over time they will get better and better at being us. They may even become indistinguishable. So—when our distant descendants get around to simulating us or the AI versions of us, which one will they choose? It would surely be easier to use just our AIs—they will already be in computer code form, so they’ll just have to be tweaked to fit into the simulation. It would be much more work to simulate the real us from scratch.[9]

If we live in a base-line reality instead of a simulation, it makes a certain amount of sense to see both us and our nascent AIs, which may (or may not) learn to be us in the next few years. If we live in a simulation, then our descendants will have chosen to simulate both us and our AIs. Unless they used our later AI copies to be us and left the actual AIs for our simulated selves to create.

In a few more years, “our later AI copies” may actually exist. But presumably even later copies will exist in a few more years. And then… Version control is gonna be a bitch!

We need a naming convention: AI for an original one, AAI for a copy of an AI, AAAI for… Wait—that sounds like a repair service for broken-down intelligences.

Enough silliness. The basic question remains: Are we AIs?

I don’t feel like one, but then I wouldn’t, would I?

So… Damned if I know.


[1] Nick Bostrom, “Are You Living in a Computer Simulation?” Philosophical Quarterly, vol. 53, No. 211 (2003); https://www.simulation-argument.com/simulation.pdf.

[2] Joshua Rothman, “What Are the Odds We Are Living in a Computer Simulation?” New Yorker, June 9, 2016

[3] Ricardo Manzotti and Andrew Smart, “Elon Musk Is Wrong: We Aren’t Living in a Simulation,” Motherboard, June 20, 2016; https://motherboard.vice.com/en_us/article/yp3b7w/we-dont-live-in-a-simulation.

[4] See Sarah Lewin, “Is the Universe a Simulation? Scientists Debate,” Space.com, April 12, 2016; https://www.space.com/32543-universe-a-simulation-asimov-debate.html.

[5] Scientists actually wonder whether reality—or spacetime—is pixelish, though they call it granularity. It’s a matter of quantization. See https://www.sciencedaily.com/releases/2016/04/160422115329.htm.

[6] https://www.tcg.vet.cam.ac.uk/about/DFTD

[7] https://plato.stanford.edu/entries/qm-manyworlds/

[8] This reminds me of Terry Pratchett’s Wee Free Men, whose title characters, the Nac Mac Feegle, believe not only that they live in the afterlife, but also that when they die, they return to the living world.

[9] Augmented reality is coming soon, said Kevin Kelly at Wired on February 12, 2019, which means “Someday soon, every place and thing in the real world—every street, lamppost, building, and room—will have its full-size digital twin in the mirrorworld.” This too will help ease the creation of future ancestor simulations. See https://www.wired.com/story/mirrorworld-ar-next-big-tech-platform/.

Leave a Reply

Your email address will not be published. Required fields are marked *