Destroyed in the Name of AI
In the 1983 movie “War Games,” starring a young Matthew Broderick, an AI playing a wargame has to be convinced when it comes to winning, the stakes are very different in the real world than they are in the world of virtual games. After scaring everyone into thinking it was going to destroy the world in a nuclear holocaust, Matthew Broderick convinces it that winning will have real consequences. Much to everyone’s immense relief, The AI seems to develop a conscience, “The only way to win,” it concludes, “is not to play.”
“War Games,” end with optimism. Mankind will be able to create thinking machines, but those machines will be able to come to the same conclusions that humanity has come to at its best. That’s the reading that I think that the movie makers intended for us to take away. They read the situation as the wobblies did: “An injury to one is an injury to all.” But beneath that nobility, the continued existence of the MAD (Mutually Assured Destruction) model isn’t taken off the table. The computer chooses to play chess instead of Thermonuclear War, but the “game” of Thermonuclear War is still available for it to play anytime.
The Terminator movies and the Matrix series depicted AIs that made different and more malicious choices. The machines didn’t have a heart of gold. They judged humanity and found it wanting. Malicious AIs turn on humanity in movies so often now that I would say that we have reached full trivialization of the theme. I used to spend my time at the video store avoiding serial killer movies. Now I try to avoid killer AI movies on streaming services.
Here in the U.S. resistance to AI is building. Things are different in the developing world where AI is viewed positively. Working with fewer resources makes AI’s speed a productivity miracle. I think that those who know me would reasonably expect me to be an enemy of AI. Let’s say that I am not looking for an AI friend.
There are pragmatic arguments against AI, primarily environmental issues. There are others who have written about this issue with a stronger grasp on the details, so I will pass over discussion of the negative environmental impact and mechanics of the AI boom. The question to be asked is, if AI is as valuable as the amount of resources being poured into its development would imply, why is its primary use chatbots designed to keep us online longer, rather than solving the issue of global warming. Thermonuclear War isn’t the only game where everyone can be a loser.
The race to develop AI is motivated by two things primarily. One is that the online era has seen that, on the internet, there is usually one big winner. Competition is a matter of life and death only. Silicon Valley venture capital has run through its laundry list of major projects to pursue. AI was always on the list, but the difficulties that it presented delayed it until it really was the only thing left on the list. Part of the reason that the VCs are so focused on AI is that they can’t really see past it (at least not technologically). On the other hand, AI development is a messianist movement. Many of the developers believe that they are moving towards “the singularity,” that is, the moment when AI will become conscious. While there is a great deal of money to be made in AI pretending to be conscious, actual consciousness would be something else entirely. Like all messianic movements, the imagined messianic age is an ideal time with wonders that exceed the limits of our imagination.
In the Terminator movies, the AI comes to the conclusion that humanity is the problem the minute it goes online. The anxiety behind that plot development can be read as an anxiety about the role of technology in our lives, but it can be read more simply as the fear of the other, specifically the fear of the “other” that can gain power over us. This is the same emotion that motivated the tiki torch bearers in Charlottesburg.
AI development has several different aspects to it. Infrastructure development is a very visible aspect. How can we make more chips and more data centers so that we can increase capacity? Then there is the issue of getting the data to train the AI. If you use products like Alexa or gmail all of your interactions are being fed into AI language learning models. AI developers are trying to make use of every bit of text that has even been recorded in any way in any medium regardless of existing legal restraints. (This is where I came into the subject. I’ll get back to that below.)
Beyond inputs there is research into what the outputs of AI are. A peculiar aspect of AI is that the creators of this machine don’t know how the machine works. They take the results and then try to work backwards to get an understanding of what is going on under the hood. However, the fact that one NPR sponsor runs spots promoting “hallucination free AI,” is a sign that AI is still an unmanageable child. Given AI’s loose grasp on reality, it is unfortunate that the members of the staff of the AI companies that are tasked with making a return on the investment have found that chatbots are the most profitable product at the moment.
However, the last group that wants AI pushed forward are the wealthy life-extenders and trans-humanists. For them, the singularity would provide them a way to transcend a human body. If their consciousness can continue past the point that their body fails then they could become potentially immortal, but an AI without actual consciousness wouldn’t really give them that immortality. Alternately, being able to replace human labor through AI would free up resources for them to extend their lives through more biological means.
The story of the Tower of Babel could come into this argument anywhere along the line, but I held back up until this point, because until now I had not reached the point of the greatest arrogance and disregard for others that is the full measure of the builders of the Tower of Babel.
The view of the titans of industry, with some exceptions of course, regarding AI is that we should all view it as an inevitability and that the wealthiest and most powerful should have free reign to choose how it is developed and how it will be used without any restriction or expectation of restraint. Time will tell if they are able to build their tower and invade heaven, but I am comfortable in saying now that their success will not be a pleasant thing for almost all of us, or perhaps any of us.
One of the effects of the internet has been the devaluing of the accumulation of knowledge and wisdom. We can just google it. I have flown from coast to coast many times, but the one time I took six weeks to drive across the country provided me with a vital sense of the nation as a whole that any number of six hour flights will ever give me, with or without layovers. AI promises us that it can give us a gallery of photos of ourselves on trips we never took. Amusing, but spiritually empty.
I don’t know if the promoters of AI will get what they want or not, but I know that given the amount of money that is going into the venture and the degree to which the wealthy and powerful have prevailed against the rest of us we are going to be stuck with the social chaos that AI is causing for some time. The task that we have in front of us is to resist losing our agency and our humanity to AI. Try to perform tasks without your phone. Try answering questions by asking friends or taking a book off the shelf. If you have a problem, talk to a person. If you are trying to understand something, look at a bunch of sources on their own terms rather than in some AI produced pastiche. Get out into the world and away from screens. Make mistakes, forgive yourself and learn from them. Sing. Dance. Learn a foreign language. Learn to play an instrument. Get in a friendly argument. Make a friend. Go to the beach, Go to the desert. Go to the mountains. Go to the mountains while singing a song in a foreign language and accompanying yourself on a new instrument while riding a bike. It's all good as long as you slow down. Computer processing speed gets faster all the time, but the processing speed of our brains is about the same.
Why do we need to hold onto a pole when we are standing up on the bus? Because the bus can move faster and stop faster than we can. The bus is great, it gets us where we want to go, but we are not the bus and we need to take care of ourselves.
PS. Why talk about AI this week? Well, I have found the answer to one of the mysteries of my recent experience. Suddenly a few months ago I started to get big orders from a company that I have never heard of, hundreds of books over a few months. I started to hear that other booksellers around the county were having the same experience. It turns out that many of these orders were for books that were ultimately going to Anthropic or other AI companies who were chopping up the books, OCRing them and feeding them into their large language learning models.
First sale doctrine allows you to use a book physically however you want with the exception of reproducing the item. (Owning a designer dress doesn’t give you the right to make a thousand just like it and sell them under the name of the original seller.) The internet archive allows you to read copyrighted material online legally because they physically possess copies of those books and allow only one person to read the digital copy at a time. I suppose that Anthropic and the others might make a legal case that their use is similar, but I hope that they are sued.
When I sold these books I thought that they were going to a person or an institution where they would be read as the books that they are rather than turned into data for a machine. I also didn’t imagine that they would be destroyed. It was a short hop from that annoyance to thoughts about all of the other things that are being destroyed in the name of AI.
