I think I realized around halfway through writing this blog that it was going to eventually end. Everything that I’ve written here is more or less my opinion (supplemented with facts) and, eventually, I wasn’t going to have anything additional to say. Which is fine. My original intention was to get a coherent idea of why I didn’t believe that the Singularity would happen. That is, I knew that I didn’t believe in it. I had some idea as to why but I wasn’t sure exactly. I think that getting down my thoughts has (with the help of commentators and others) taught me what it was I believe in. Which is useful. To me, at least.
The Singularity is a relatively esoteric and geeky topic. I don’t currently have a lot of friends who even know what it is, let alone those who would argue with me over its details. So I turned to the internet. Here I can share my ideas and get feedback and criticism and work on refining my own ideas. I want to thank anyone who has been reading this or who has offered their two cents. My aim was to generate a dialogue and stimulate thoughts.
In a sense I’m very glad that this is over. Judgments and skepticism about the Singularity have not stopped plaguing me for over a month and I will be glad to take a break. I will have time in my head to think about other things. More important things.
And maybe I’m still stuck in an older paradigm of writing. I like books. I like the way they are finite, they are self contained. The fact that they have a beginning, a middle, and an ending. So I like treating this blog somewhat like a book. It’s all on more or less one topic. It’s bounded in time. I’ve don’t think I’ve ever sat down and written so much about just one idea; this blog probably runs between twenty-five and thirty thousand words. For me, that’s a lot. In fact, it’s enough.
In another sense, this is nothing like a book. You don’t have to read it linearly, you can start and stop at any one post. You can read as much of it or as little of it as you want. It’s full of links and commentary and incorporates pictures, video, and supplemental material in a way that a book never can. And it’s all a rough draft. Most of the time, when I write something here it’s the first time I’ve written it down. But that’s okay. It’s here to be judged and criticized, that’s why it’s available to all. So I like having this online.
Reading back over my posts, I tried to see if I really did come to some conclusion with my thoughts. Somewhat, I guess, but not entirely. I don’t have everything covered. There’s plenty of more detail I could find. There are larger topics at hand. There are holes and probably some inconsistencies. But that’s to be expected. I’m just a man, it’s the best I can do. And I think that I came to some sort of consensus on why I don’t believe in the Singularity. Here’s my own summation:
1. People tend to stretch a single idea too far. The idea that processing power is everything, that once we get the equivalent processing power as the brain then AI will be inevitable, seems like a part of this. Processing power is one thing. A modern off-the-shelf calculator has the same processing power as the Apollo spacecraft and yet doesn’t routinely fly to the moon and back. For that to happen you need lots of other things: a well engineered spaceship, a huge amount of money backing you, the political willpower and climate to make it feasible, trained astronauts, an idea of spaceflight set up in the culture, a scientific rationale for going, the blessing and dreams of millions of people who believe it to be necessary, an economic incentive, etc. All of these things (and more) were in place in the 1960s to make space travel possible. Similarly, there will have to be a convergence of more than just technology in order to make AI possible. There are many ways for it to come about. And people like Raymond Kurzweil and Vernon Vinge are engineers. They have a great deal of authority when it comes to technology and engineering. I trust that plenty of Kurzweil’s purely technological predictions can come true. But he downplays or forgets the importance of other factors and so I think his ultimate thesis is wrongheaded. Based purely on technology, we could have colonies on Mars right now. We probably have been able to for twenty or thirty years. And yet, it hasn’t come about. So might it be with AI.
2. The mind is like a computer but it is not a computer. Yes, it works on mechanical, physical principles. But that doesn’t make it identical to a computer. Giving estimates of the “processing power” of the brain is largely going to be guesswork. It’s an estimation based on a metaphor based on a particular view of the universe. Furthermore, the brain is an integrated whole that is integrated into a larger whole. It’s hard to figure out the levels that the brain operates on because everything inside of it is interconnected. It affects itself affecting itself. And it gets affected by the larger organism that it inhabits. I am not just my thoughts nor just my brain. I am me, every part of me, down from my wiggling toes to the split ends of my hair. And I am in constant flux. At any moment, I am taking molecules from my environment and incorporating them into my body. Every cell wall that is a part of me is a jiggling mass of continuous change. I am a standing wave of barely controlled chaos traveling through space and time. Recreating that is hard. In order to simulate a person, we will have to simulate both hardware and software and the fact that the two are really just one. You will have to account for the fact that I am just a collection of atoms borrowed from the universe for a span. I don’t know if perfectly simulating that is possible. Perhaps a different sort of mind is. I believe that we can get AI, it will just take a long time and it will come piecemeal.
3. Intelligence is not a “real” thing. We like to think that intelligence has to do with getting a high SAT score or being good at chess or being able to answer trivia questions. But those are all just views of intelligence. Smartness is more like beauty. We can largely agree on what is beautiful and we can even “enhance” our beauty with modern surgical techniques (but that doesn’t lead to a “beauty explosion” of ever more beautiful people). The best we can do is make human beings more beautiful based on some pre-existing idea of beauty. They start to match up closer to that idea but, ultimately, it’s just vanity. Maybe someday in the future we will discover techniques to “enhance” our intelligence in the same way that we do plastic surgery. But we will have to work with what we’ve got. There isn’t some switch or dial that you can flip to make a person more intelligent. And yes we are very intelligent, in some ways more intelligent than all other creatures on Earth. But all organisms are problem solvers. They model their external reality just as well as every other creature around them, regardless of whether or not they are bacteria, fungi, plants, or primates. If they couldn’t, they would be eliminated. We can design tests of intelligence but they have to match up with some pre-existing idea of intelligence. All they can ever do is show how one person or organism does better on that test than others.
4. Knowledge is wonderful. But in and of itself it can not accomplish the impossible. It can not make a perpetual motion machine. It can not stop entropy. It can not preclude the fact that a wrench will be thrown into even the best laid plans. The universe is not a series of stumbling blocks that we are in charge of solving. We are taught to think this way in our high school classes: find the hypotenuse, derive Kepler’s Laws from Newton’s, tell me how Coleridge and Wordsworth were similar, give me the causes of the Civil War. But we use those illustrations precisely because they have answers (some of them, at least). Life is not a question and answer session. Real problems, the kinds outside of textbooks, don’t have “real” solutions. Even those in physics and math. Answers always end up generating more questions. Usually three or four. Our knowledge can be great and amazing. But nonconscious and unconscious forces are often greater and more amazing. I still trust evolution to manage things much better than we can for the time being. We are still too stuck in linear thinking; that problems only have one cause and one solution. That DNA controls cells. That pills can target specific emotional states. That robots will either love us or hate us. Reality is made up of infinite shades of gray and just thinking about things will not generate a remedy for everything that ails us. The universe does a much better job of shepherding itself because it does not have to deal with questions and answers. Our way of thinking can accomplish many things. But a nonconscious entity does not have to couch itself in this “problem and solution” view. It has a different way of thinking. Perhaps, if we do eventually create AI, it would be smart to make it nonconscious.
5. We do not occupy a special place in time. We already know that the Copernican Principle says that we do not enjoy a special place in space. Why do we insist on thinking that our time is the most important time of all? For most of Western history, people have thought this. They said, “Now the Messiah will come. It will happen in my lifetime.” We don’t think like that anymore but we hold on to certain ways. There is not some special point in time when everything that comes after it will be different (or, to put it a different way, every point in time is such a time). We are not living at the height of civilization nor are we at the lowest point. We are at a point. To insist that things are inevitable, that there is some destiny in the universe, that things are not contingent, and that this is the only way that could be is, to me, an extremely old-fashioned way to think. I always thought the point of atheism was knowing that there isn’t some design or plan in the universe. To say something is inevitable contradicts this, in my opinion. There are patterns. There are interesting places and interesting times. But we are the ones that decide what is interesting. We are the ones who see the patterns. Destiny is in our heads entirely. We are the ones who shape it for ourselves. We have to deal with that.
In and of themselves, do I think these points are enough to say the Singularity won’t happen? No, of course not. These are my thoughts. But there are counter-arguments to nearly every argument. Raymond Kurzweil spent a chapter of The Singularity is Near attacking some of the points I just raised (though not well enough to dismiss them, in my opinion). Furthermore, the Singularity is a very powerful meme. I suspect that it will last longer than Kurzweil and Vinge, longer than 2045, longer than plenty of us. People will forget what we’ve already said and then say, “No, really, this time it’s just around the corner. We have better facts, we can prove it more fully now.” It’s what always happens. So I don’t expect this debate to go away. I’ve said my peace about it, for now. Perhaps in a few months I’ll come back and start revising and cleaning up some of these arguments. In the meantime, I hope that there is more debate and discussion.
I just want people to question the steps that occur between the steps that are given for the Singularity. When someone says, “We will build a smarter-than-human intelligence and then the human era will be ended,” I have to question that. What are the steps between those two? More importantly, what are the steps required to get to the first one? Every single one of those steps (and there are a lot of them) could be proved wrong or have to be altered with new information in time. I feel like so much of the discussion that I have seen about the Singularity assumes that it has already happened and now we can elaborate the detailed minutiae of what comes after. But what about the detailed minutiae of what comes before? In my opinion, a lot of the writing about the Singularity tends to focus on the bombastic and misses the subtle. I am a big fan of subtlety. I wish that there was more incorporation of it into the discussion. Moreover, the analysis seems to focus on a handful of scenarios (whether AI is friendly or not, etc.) out of the millions possible. To have the discussion based on the assumption that those are the only possible options seems to disinclude plenty of alternatives and choices.
That being said, almost every single person who has answered my questions or written here in the comments has been incredibly reasonable and skeptical. I think I originally had some naive view that people who believed in the Singularity needed a reality check. I am glad that this has been overturned. The people I’ve met through this endeavor who believe that the Singularity will happen have all been gracious, intelligent, commonsensical, credible, and keenly aware of the limitations of their own arguments. Because of their persuasiveness, there have been plenty of times when I was able to see through their eyes. I have tried to emulate them. I want to give a heartfelt thanks to those who engaged with my (possibly frivolous) objections.
As to what ultimately will happen, I haven’t got a clue. I have suspicions. But I guess we’ll all just have to wait and see. I can trust that, as the future is not yet written, all of us can expect to be surprised. I have hope that we will all be pleased, as well.
~ Adam Mann (07/05/09)
