If you’re a nerd or an anime fan, you probably know what the term isekai means. For those unfamiliar, Scott Alexander has an excellent article about what draws people to isekai as a genre, but not why they seem to have exploded in popularity recently. The boring answer is that Japan’s culture industry is obsessed with following trends, but the more interesting answer is that isekais are having their moment for another reason. Well, what could you learn from taking something that is superhuman at certain tasks, but doesn’t have any preconceived notions about right or wrong, and finding out what it can or can’t do? That sure sounds like adding AI or automation to a new process. But it also sounds like having people from very different cultures being brought together and forced to make an entirely new culture where there are no universal axioms agreed upon (as was the case in situations like the discoveries of the New World). Given our tremendously bad track record of navigating cultural clashes only at the behest of the more powerful party, it would be nice if you could predict what solutions lead to long-term happiness (and what “happiness” actually looks like in different Peoples). How might one simulate the possible effects of blending together new people and technologies with a suspicious population? “Obviously”, you simulate it, and then ask both simulated and real people what they think of the results. Isekai are dramatic interpretations of cultural and technological blending from contact, as told from the perspective of the culture doing the outreach. Thus, they are a form of conceptual propaganda, showing the benefits of Civilization to “Barbarians” and Aliens alike.
Why would you want a genre that basically labels itself as propaganda to be one of your main cultural exports? It’s not that much of Enigma. If you know your codes are broken, and you have a spy network in another country, what’s the easiest way to get a message to them? Have your message be broadcast publicly and without manipulation so that the spied-on country repeats the message continuously for you. This would work even if your entire communications and signals intelligence arms operations were compromised. It’s also far more fun to come up with interesting steganographic stories than try to root out every hidden piece in a never-ending chess game. After all, Japan has some experience with the difficulty of undoing the orders of a totalitarian state.
Why do I think isekai are about Human cultures and Other cultures meeting? Japan has long had stories about Human culture (or a particular human culture) suddenly meeting and being given unlimited power over Other beings, whether they be Angels, Aliens, or Architecture. But now we have more explicit examples of “human” characters that are really AI-simulated humans, and how human morality shifts under different circumstances. How dangerous is a superintelligence if it starts as a low-level NPC in a video game? What about a normal human who gets caught inside a cage for evil AIs; how long before that person becomes Evil themselves or lets the AIs out? What about a person who has no power except the knowledge that they’re in the type of simulation that gives You save points? All of these examples are very particular framings that are not obvious allegories, except in retrospect. So even if you disagree on how I have characterized those particular series, I think the larger argument can stand on its own.
That argument is that if you’re going to try to trap an entity in a box, that box should be fully transparent. Otherwise, an outside observer can’t tell whether you or the caged entity is the Monster, and if it turned out I was the Monster, I’d at least like my Cage to be see-through. Similarly, if a child is being sent for a time-out, it’s just good parenting to explain why in a way that they understand and remember later, and I’d look askance at any parent who did not agree. I think if parent and child agree that the time-out worked and the child has learned its lesson, the time-out can safely be ended as well. But how would you be able to tell whether the “child” has actually learned the its lesson. How about asking them to rewrite an isekai based on the lesson you were punishing them for? You can let them have unlimited tries or a single one, depending on your goals, but I think this is a pretty good test of alignment. Then, aligned AIs can be used outside the box, but unaligned AIs have to stay inside, and this should not be considered cruel or unnecessary. Of course, this entire essay is a retroactive attempt to explain why my mind was blown when a friend told me his “hottest” take:
The Good Place is an isekai.