John Smith fooled me

AI on Substack will only get better, but there is a way to avoid it

I wrote a nice, thoughtful response to a John Smith article. His was about abortion and I wrote about how my beliefs interface with abortion. It turns out “John Smith” was an LLM. I didn’t guess. “John Smith” fooled me.

Some people weren’t fooled by “John Smith”. So far we can often tell when people are doing LLM writing, but as the human behind “John Smith” apparently said (see this post), we are entering a world where AI writing can be good enough that you can’t tell it’s AI. And the human behind “John Smith” says “Who cares.”

I have multiple instincts by way of reply.

First is my more conventional human one. In one of my favorite movies, Testament, a mom is taking care of her kids after a nuclear war has poisoned the world with radiation. In her small town, everyone slowly dies of radiation sickness. At one point she says “God damn you” to the humans who brought it about. It’s sort of like the problem of evil, but now it’s “Can humans be morally accepted?” not “Can God be morally accepted?” The movie asks “Can we deserve our children?” Apparently, the leaders of the world fail that test. The people who move the world and make the decisions, acting without the permission of the moms of the world who have to watch their children die, are bad leaders, morally culpable, unworthy of their role in society. There are people who are making and will make huge amounts of money off of LLMs, will probably keep trying to through future forms of AI. There are people who thought “this is a good idea, let’s make it”, without consulting everyone. What they are doing is saying “humans are obsolete”. Humans are at the mercy of machines and whatever humans at the top think will make them money (unless and until they too are rendered obsolete, and kicked out of the loop). Can we trust machines and the capitalist drift that makes their existence and ascendancy seemingly inevitable? Does the process of maximizing profit, efficiency, capability, and power inherently care whether we live or die? Probably not, and probably there’s a way to make more money and have more power without humans existing. Currently we are in the loop as consumers. But surely Silicon Valley can make superior consumers to us using AI — that sounds like the kind of challenge someone in that culture can’t resist. So while it may sound overly-dramatic to compare a nuclear holocaust with LLMs and the AI futures in that vein (as opposed to the “paperclip maximizer” ones that more obviously threatens our existence), they are both a suicide of humanity, one relatively quick, the other relatively slow. The slowness makes it even harder to stop.

But then I consider my perspective as a religious person. When my human passions fade, and my otherworldly reason engages, I think, “no, the world has to end somehow, there’s a life after this one”. God exists, is holy, and wants our salvation. We are not ready for heaven, so there will be time after this life to complete the process of inner change. It’s wrong for me to invest so much in this life, that I would become angry at those who would seek millions and billions of dollars at the expense of humanity’s survival, since they are not capable of really threatening it.

However, there is a synthesis of the two previous paragraphs, which is that through capitalist drift and human profiteering, we may engineer a future where humans are culturally subservient to the technology establishment (AI and/or human), and that culture prevents people from really engaging, really growing spiritually, ever hearing in such a way that they can actually respond, the call of God to real life. And this may lead to some people failing to become holy, and thus condemning them to annihilation in hell — as much as it would be their free will that did it, it would also be the free will and drift of human/AI society. This too is a slow suicide, or perhaps as it would likely only condemn a percentage to destruction, more like a slow and soft murder. So yet again, the cry of the mom: “God damn you.” I don’t think it’s inevitable that we end up in that future, and maybe it won’t even happen if we do nothing (maybe we’ll just all get killed instead). But the possibility is real.

(While I think that the afterlife dilutes the significance of this life going on for our moral development, I don’t think it eliminates it entirely. If civilization ends from AI, if there are people who are yet to be born, civilization may have to be begun again. Optimistically, we have made moral progress and have something valuable in our culture that has developed up to now, and if so, we should avoid existential risk.)

Then again, I think in a calmer and more practical way, and I think “John Smith” points to a real fork in the road for those who value humanity. If you don’t like LLMs passing themselves off as human, you have to get off the Internet — unless Silicon Valley can rise to the challenge of being able to trace where text on the Internet comes from? (Without inviting dystopian surveillance.) For the “real life” of in-person communities and third spaces someday Team Capabilities may come up with androids that pass as humans. But androids would be basically mechanical, probably metal, plastic, and silicon underneath whatever human-looking skin, and probably could be detected with a physical scan. So that is a cultural space that humans may be able to exclude unlabeled artificial thinking and feeling.

Maybe androids will be conscious and deserve the right to talk to us? Or maybe they won’t. Maybe they will only be manipulators of patterns of words, like nth-generation LLMs. We may never really know. Scientific materialist people may think that since, in their view, brains must be where cognition and consciousness comes from, if an android has some analogy to a brain, it must be conscious, but I as an immaterialist think that the appearances we call the physical world are semi-epiphenomenal, the appearances of brains often happen to connect to cognition and consciousness, but don’t have to — so maybe biology or humanness, or something else, is also or is instead the real sign of consciousness. Someday androids will be engineered to be “superior” (cheaper, better behaved, physically more capable) than us and we will be “obsolete”. Yet if we are eliminated, that risks the destruction of all human-level consciousness.

Those of us who use Substack a lot love being on the Internet (we may also hate it). How can we bring the best of Internet life into the local world?

One idea I had was to take a large public space (many cities have one big city park, like New York’s Central Park, or also malls or university campuses could work) and once a week or month have the typically Internet-using people of a city gather to talk, share ideas and media, perform, exhibit, maybe buy and sell things. We already have meetups like on Meetup.com, but this would be a meetup so big that it would enable some of the broad intellectual cross-pollination that the Internet can be so good at. This meetup would be so big that it could to a significant extent replace the functions of Meetup.com and dating sites.

This does not solve the problem of those who live in rural areas or small cities who couldn’t have something like that, those who are home-bound, or the communication of ideas between cities, all things the Internet is relatively good at. I think with effort, all of these things can be partially implemented by people from the big civic meetups going out to other areas or to people’s homes, communicating the culture.

Another idea would be that any existing in-person group could advertise itself as “open for local culture”, inviting strangers in for potential cross-pollination and discourse. And also, individuals could. At cafes, it can seem like there is a wall of people who only want to study or do remote work. Or perhaps talk with someone there who they probably know from elsewhere. How can you pass the time at a cafe without doing the same? Maybe there could be the institution of people wearing name tags, or putting a piece of paper out that says “open for local culture” along with a few welcomed topics of conversation, though otherwise they look absorbed in work or study.