A Critical Examination of The Singularity

the technological singularity. I've done an episode about the singularity before, but I think the last time I really talked about it was in and I think attitudes have changed a bit in the five years since I did that episode. So I have argued and will continue to do so, that the technological singularity is still a very vague term. It means different things to different people. It can even mean different things to the same person depending upon the situation.

In fact, if you'll allow me a little allusion to literature, since I do have a background in scholarship and English lit that I rarely get to use. This actually reminds me of Humpty Dumpty and Alice through the Looking Glass Humpty Dumpty in that talks about how words only mean whatever he wants them to mean, and all it takes is the will to use a word however you want, and the word has to obey. That seems to be

the general feeling around the technological singularity. But if we're to explore this concept, we're going to have to define it at least a little bit. So here goes my attempt. First, let's go with the singularity in cosmology, right, what does it mean in outer space? What is it in physics? In physics? In outer space, it refers to the one dimensional point that is at the very center of a black hole. This is the point where density and gravity

are infinite. It's the point where spacetime curves infinitely, and most importantly for the concept of the technological singularity, this is the point where the laws of physics no longer apply. They break down. So the technological singularity is ultimately a metaphor based off this cosmological notion. So with the technological singularity, the idea is that we arrive at a moment where technology is able to iterate and evolve at a rate so fast that it is impossible for us to envision

or predict what happens next. Or it's about superhuman intelligence. Those tend to be the two big ones. So generally we imagine that superhuman intelligence actually brings about this era of rapid technological iteration and evolution, and the nature of that intelligence is still a matter of conjecture, because as far as we can tell, it hasn't happened yet. Some days we struggle for normal human intelligence. But there are some big, broad categories that we can talk about as

a possibility for this superhuman intelligence scenario. There's the artificial intelligence version, in which computer scientists design a system of such sophistication that it operates on a human like level of intelligence. This AI, in turn begins to design its successor, which is an attempt to improve upon the previous generation.

This next AI ends up being smarter, faster, stronger than its creator was, and it goes on to create the next generation of artificial intelligence, which by definition is going to be better designed than anything we humans could accomplish, because generation two is already better than human level intelligence, and bam, at that point you have the intelligence explosion. You have a superhuman intelligence beyond any capability of human thought here and because it's operating at a level beyond

human capability. We cannot possibly guess at what will come next because we are only human. We can't conceive of what will happen with superhuman intelligence because we don't possess that. The vision of the future is sometimes but not always, accompanied by the notion that these machines are going to gain sentience, that they will have some form of consciousness, and therefore they are consciously making these improvements so they'll

have some sort of sense of self. That also suggests that there's the possibility that they might develop motivations and wants and needs. So this leads us to some scenarios where this all develops into a terminator or matrix like dystopia, in which these machines have come to the conclusion that we human beings are a blight upon the earth and

that we have to be wiped out. Or it could lead to machines subjugating humans in some way, that we are the ants to its colossus, and we can either be ignored or we can be put to work in some menial task. Or maybe the machines would just be super nice, high tech guardian angels that are granting us wishes whatever we want it they make it possible because

they're intelligent enough to do it. But the point is the machines are the ones making the decisions, and they would be doing so beyond what we could do ourselves, and therefore we have no way of predicting what would come next or what the future would be like, because it's by definition, outside of our own experience. Another version of this superhuman intelligence that will bring on the technological singularity comes about due to biological or technological boosting of

our own intelligence. So instead of making computer super smart, we make ourselves super smart. Possibly with the use of computers, we've discovered a way to augment our intelligence beyond what was previously thought possible. Now, in a way, you could argue that the history of humanity has been one of using technology to augment our intelligence, though the ancient philosophers

like Socrates would disagree. Socrates would say, don't write anything down, because if you write things down, you don't have to keep that information in your head, and if you let information out of your head, then you're less intelligent as a result, you didn't have to keep it of there

in your noggain. Similar arguments were made in the wake of the Guttenberg printing press when that was invented, and more recently, much more recently, Nicholas Carr wrote about the same idea, only his target wasn't the book or the printed word, it was Google. He wrote, is Google making us stupid? So since we started writing things down, there have been people who have said this technology is going to make us all dumb, because you don't have to

think if you have this stuff to rely upon. But the counter argument is that the technology can augment our intelligence. We can offload some of the cognition onto the tech or that we're carrying or that we're depending upon, and that frees up more of our capacity to reflect or to do free association of seemingly separate and distinct ideas to bring them together there and come up with innovative

solutions to hard problems. So I think it all just depends upon your point of view and your approach to using technology. Anyway, in this vision of the singularity, human beings become endowed with superhuman intelligence, which might very well mean we would no longer really be what you and I would consider human at this point. This is the

trans human vision of the future. So the end result would again, be that we're talking about reaching a point where we, as humans, flawed, limited in our intelligence, are incapable of imagining what would come next because we cannot put ourselves in the place of a superhuman intelligence. We lack that intelligence, whether it's machines or us, or some combination of the two. The basic idea of the technological singularity hinges upon the emergence of this superhuman intelligence in

almost every case. So where did that idea even come from? Well, back in seven there was a publisher named the Reverend Richard Thornton who wrote a piece about a mechanical calculator that illustrates my earlier point about how people have talked about how technology can make us dumb or it can be a hindrance for thinking. He wrote that such a device, if placed in a school, would do quote incalculable injury

end quote. Don't think he was necessarily intending to make a pun about a calculator, but he did go on to say that these machines might get better and better, and what might happen next? He said, maybe they would be able to calculate ways to improve their own design and quote grind out ideas beyond the Ken of mortal

Mind end quote. So he's saying, these machines themselves might even come up with the method of making a better machine, which in turn might make even better ones, and eventually get to a point where they could produce information that we humans can't even dream of. This is back in

eighty seven, before the electronic computer. Alan Turing, whom we all often associate with this idea of machine intelligence, wrote a paper in nineteen fifty one at which a hypothesized that once machines do level enter a level of like human like thinking, or at least appear to think like humans do, it wouldn't take long before they would outpace

our own capabilities. And he cited the nineteenth century writer Samuel Butler in this Butler himself had said there is no security against the ultimate development of mechanical consciousness in the fact of machines possessing little consciousness now, essentially saying, just because machines are dumb right now doesn't mean we can discount the possibility that they will one day be intelligent.

He was making an observation about how quickly machines were advancing, and he was comparing that against the very slow process of evolution in biology. He was saying, you know, it took millions of years for lower organisms to evolve two higher thinking organisms. But look how quickly technology is developing. It won't be long at all before those are thinking before machines can think. That seems like a somewhat faulty

reasoning to me, but we'll move on. John von Neumann, who I talked about in episodes in Tech Stuff this year, reportedly talked about a singularity that would happen once technological process a progress, rather would occur so quickly and become so complicated as to become incomprehensible. So this is more in that technology is rapidly iterating and evolving, but not

necessarily related to a superhuman intel aligence. Uh. He told his buddy Stanislav Ulam about this, and it was Ulam who reported the conversation, so it's more from a secondhand information. Then there's I. J. Good, who was one of tourings peers at Bletchley Park when they were working on cryptography during the World War, and he wrote about a future intelligence explosion. He described the scenario that I mentioned earlier of machines designing ever more capable successors and leaving human

ingenuity in the dust in the process. His work would inspire a science fiction author named Verner Venge, who presented the earliest recorded version of the phrase technological singularity. He wrote it in a piece that was published in Omni magazine in nineteen eighty three. A decade later, Venge predicted, quote, within thirty years, we will have the technological means to create superhuman intelligence. Shortly after the human era will be ended.

End quote sounds pretty ominous, but honestly, it could just mean that we humans would be evolving into something new. Now, that was a prediction he made in and he said it was thirty years in the future, which means we're coming up on that deadline awful soon. It would mean that we would be there and let's see a two thousand twenty three by my math. Venge also suggested four

ways this singularity could emerge. Computers could reach a certain level of sophistication and become awake and you know, therefore conscious and they would be superhumanly intelligent. Or computer networks would become suitably complex enough for intelligence and consciousness to emerge from that complexity, so the network itself and the associated users would become kind of a collective superhuman intelligence.

There could be a computer human inter face future where these interfaces are so sophisticated that we can augment our intelligence seamlessly with machines, and together we are superhumanly intelligent. And finally, he said, we might figure out a way to biologically boost our own intelligence. Now, I would argue that Venge is one of the two really famous champions of this concept of the technological singularity. The other one I will cover in just a moment, But first let's

take a quick break to thank our sponsor. I think the person who we most readily associate with the concept of the technological singularity, he has written numerous books on it, and has given lots of talks on the subject, and has promoted the idea extensively, would be Ray Kurtzwild. Kurtswilds definition is a little different from Vinge's definition, and the singularity is near. He talks about the era of rapid technological development that will make defining the future or even

the present at that point impossible. And I think this is closer to what Stennis law Oolam said John von Neumann had told him. Kurt Swell, I should point out has made numerous contributions to technological development. I don't want anyone to think that I think Kurt Swills a crackpot. I don't. He was a pioneer in computer pattern recognition and speech recognition. He did a lot of development and other areas, including electronic music. So kurts wells a really

smart guy. His view of the singularity tends to include some pretty far out ideas, however, including the possibility that through this process we will discover some means of extending our lives indefinitely, including the possibility of porting our consciousness into technologies somehow, in other words, finding a way to take what makes us us and put that us in some sort of machine, and that would effectively allow us to live forever, though I don't know what sort of

experience that would be, like, would you just be a consciousness inside a computer? And if so, would that be a satisfying existence? Also, there's no way to even guess how this would be achieved. Uh, I'm not entirely clear what he's basing these ideas upon in the first place. It seems to me, and this is just my own opinion that some of Kurt Swill's notions are based upon a general fear of death, and that the singularity is

viewed as a sort of escape mechanism from death. It's almost wishful thinking, in other words, that maybe once we hit this point, will have conquered death itself and we will never have to worry about facing our own mortality. I still don't understand how this sporting of consciousness would work.

Or maybe you would just make a copy of yourself, in which case you would have a digital version of you, and in the meat version of you, but the meat version of you was still gonna die, which means, yeah, there'll still be a version of you hanging around, but it's not the you that's experiencing this podcast right now, unless you're hearing this post singularity and you are the digital version. Where did you get this show? Okay, I'm sorry,

I was losing my mind there. Again. While I think that his ideas are far fetched from a personal standpoint, there is no doubt he is a billion times smarter than I am, so I acknowledge I could very well be in the wrong here. It just my my critical thinking and skeptical nature are really starting to raise some red flags. Now. There are a lot of other people who are also important in this field. I don't mean to suggest that I have covered them all. There are

proponents and there are skeptics. For example, there's Hans more Effect, who's a computer scientist and a roboticist famously working with Carnegie Mellon University. He generalized Moore's law to a broader application of artificial intelligence and the arrival of a super intelligence. And one of his predictions was that by twenty forty or so, robots will become so sophisticated at that point and so complex that they will effectively emerge as their

own species. In he penned the paper titled the Age of Robots, and this is a quote from that piece. Computer less industrial machinery exhibits the behavioral flexibility of single celled organisms. Today's best computer controlled robots are like these simpler invertebrates. A thousandfold increase in computer power in this decade should make possible machines with reptile like sensory and

motor competence. Properly configured, such robots could do in the physical world what personal computers now do in the world of data. Act on our behalf as literal minded slaves growing computer power over the next half century will allow this reptile stage will be surpassed in stages, producing robots that learned like mammals, modeled their world like primates, and

eventually reason like humans. Depending on your point of view, humanity will then have produced a worthy successor, or transcended inherited limitations and transformed itself into something quite new. No longer limited by the slow pace of human learning and even slower biological evolution, intelligent machinery will conduct its affairs on an ever faster, ever smaller scale, until coarse physical

nature has been converted to fine grained, purposeful thought. Now, his ideas are predicated upon the assertion that consciousness, which is a quality that's devilishly difficult to define. In fact, I would argue it's just as difficult to define as the term intelligence. He argues, this arises from the material that is, the mind is totally the product of our nervous system, or, if you want to be a little more generous, the combination of our nervous system and our

interactions with our environments. So, in other words, consciousness emerges from a system if that system meets the physical criteria. If true and I happen to believe that this is true. That it then stands to reason that if you have a sufficiently complex system with powerful enough machines, we should

be able to create an artificial entity that possesses consciousness. If, however, consciousness arises from some other scientifically undiscovered or even undiscoverable quality, then it wouldn't matter how complicated we build our toys, they would never become conscious. So, in other words, if consciousness were the emergence from some other thing that science cannot address, like a soul, for example, then there's no way that we could create a conscious artificial being. We

can't create the soul. If that is in fact how it works. I personally feel that that's not the case. That our consciousness does arise from the material that it does come from our nervous system, the complexity and the electrochemical processes of our nervous system. The question I have is whether or not we will ever be able to replicate that in an artificial system. Not saying that it would be impossible, just wondering if we will ever figure

it out. It remains an open question. Nick Bostrom, who served as the director of the Future of Humanity, Institute has written extensively about trans humanism. I talked about that second ago. That idea that we transcend being just humans through some process. Whether that means a computer augmented person or a biologically augmented person isn't really important, at least from this perspective. It's very important from an ethical perspective.

But he's using trans human to indicate that this describes someone who has moved away from what we would define as being a human being today. And like kurtswhile he has hypothesized that the Singularity will bring along with it some means of extending our lifespans indefinitely, but he feels that some of the more aggressive predictions are a little

too optimistic. He has said that he felt that there was a less than fifty chance that we will have developed any sort of superhuman intelligence by the year twenty three. He thinks it's going to happen, but it might take a bit longer than that. Some of the people who believe or have formerly believed the Singularity to be around the corner aren't convinced it's necessarily going to be good

for us. Venture capitalist Bill Joy, who co founded Sun Microsystems, has expressed concerns about it, and it wouldn't necessarily take a superhuman AI to do damage to us. Joy has pointed out that technology tends to advance our capabilities in all sorts of areas, including destructive ones. In other words, it gets easy you're and easier for smaller and smaller groups to do more and more harm. And that's not true of all technologies, obviously, but that technology does enable this.

So as technology gets more sophisticated, the potential for a person or even a small group of people to exploit it in order to do a lot of damage also increases. We certainly see the potential for this in some technologies. For example, we've already seen three D printers that can

be used to create untraceable firearms. But Joy's concern is that we could have a much more critical existential threat as a result of these technological developments, specifically in things like bleeding edge technologies like nanotech or biotechnology, and he feels that we need to build in the systems or create social pressures to prevent those things from happening. Similarly, Alizer Yakowski has advocated for the development of standards and

algorithms to guide AI towards a benevolent mindset. Now I imagine he's an advocate for computer scientists to include ways for humans to see how machines are arriving at answers or calculations, which I agree with. I think that is very important. Transparency is incredibly important with these machines and systems. One of the dangers I see with machine learning is an approach that would block off this process from view.

We wouldn't know how a machine got to the result it got, and we would just be taking it on faith. This would turn machines computers into black boxes. They produce results, but we don't know how or why. Such a thing would be more of an oracle than a computer, and

that's not terribly healthy. I think it would be better if we build machines that are capable of showing their work as it were, to explain how they arrived at a particular conclusion and based upon the input that they received, and that way humans could verify that the decisions the computers were making were actually sound and appropriate and not mistakes. Next, I'll talk about some of the objections to the concept of the technological singularity itself. But first let's take another

quick break to thank our sponsor. So what do people who think the technological singularity isn't going to be a thing, or at least not a thing that's going to happen anytime soon. What did they have to say? Well, one counter argument is that futurists and singularity enthusiasts are far too quick to apply the results of Moore's law across the board to all sciences and technologies that would be necessary in order to bring about some sort of superhuman intelligence.

And this to me seems like a pretty solid count her argument. So, first of all, Moore's law is not really a law, right, It's an observation. It's an observation and a prediction that Gordon Moore made in the early days of silicon based transistors. Namely, more stated the due to economic demand for electronics, the semiconductor companies would invest

the resources needed to create more complicated chips. And now, at the time he was making this prediction, he was specifically talking about how many discrete components you could fit on a square inch silicon wafer. Essentially, this was really about how many transistors can you cram into that space.

Demand would create the need to innovate. The economic incentive to innovate, so companies would spend money in research and development, and as a result, they would shrink those components down so that you could fit more of them on the same amount of space and have more powerful electronics, which would continue to create more demand. He traced this pattern back, and he saw that it seemed pretty steady, and he

predicted that it would continue. Now these days, we tend to say that every eighteen twenty four months you see the processing power of microchips double. But keep in mind that's a modern interpretation that is an evolution of the original observation about transistors. So Moore's law has changed over the years. Anyway, there are lots of reasons to reject any predictions that are predicated upon applying Moore's law to

all technology across the board. First, even for transistors, Moore's law has largely been something of a self fulfilling prophecy. Companies have acknowledged the prediction, and then they've worked really, really hard to keep up with it. So in a way, the reason we see such rapid progress in microprocessor technology over the course of the last several decades is because these companies have a sort of sense of obligation to

keep up with Moore's law. If they didn't, it would see it would be like a failure, right, It would seem like they were the ones who broke More's laws. So they keep working super hard to keep that going. And that has meant that we have reinterpreted Moore's law. That's why we think of it more as processing power these days and not in the number of transistors. But we know this is not sustainable indefinitely not based on

our current microprocessor architecture. At least, we will hit physical limits based on those designs that eventually we will not be able to engineer around if we stick to that architecture. Now, that does not mean we won't find alternative approaches. We might find alternatives. However, that might also mean we won't see progress continue at that same rapid pace. We might see improvements, but it might be on a slower scale,

which means we don't advance as quickly. As it stands, we've seen more advances in processing by changing up the way we do the work than we do with the hardware over the last couple of years. In some cases, that's by going through parallel processing either by having multi core processors or even leveraging graphics processing units g p

us to do some of this work. Now, that's one objection to Moore's law, but another is just because of general observation applies to one part of technology, doesn't mean it applies to all parts of technology. That would be crazy. Let's say I took More's law and I tried to apply it to cars, and I said, the top speed of the model of this car, model this car that I just bought, this top speedies a hundred twenty miles

per hour. I'm gonna buy one in two years. I'll buy a brand new one in two years because Moore's law states it's gonna be twice as powerful. In two years, it's gonna go two hundred forty miles per hour. Two years after that, it's gonna go four hundred eighty miles per hour. Well, if I told you that, you would say you're nuts. That's not how it works, and you would be right. That isn't how it works. The mechanical performance of engines does not follow the progress of Moore's law.

Moore's law does not apply to all technologies, evenly across the board, So we can't say that the rule, and keep in mind More's law isn't really a rule. The rule for microprocessor progress doesn't necessarily apply to other sciences and technologies, and a technological singularity dependent upon the emergence of a superhuman intelligence is going to depend upon a lot of different technologies, not just raw computing power. So

that's one big objection. There are others as well. So many of these predictions assume that we're going to see computer scientists develop software that can leverage the hardware allowing

a superhuman intelligence to emerge. Jared Lanier, who is a pioneer in many fields of technology like virtual reality, has said that there's been no evidence so far that any human programmer is capable of building something like that, and ultimately a human programmer is going to have to build the software that will enable hardware to get to the point where consciousness will emerge. I mean, if I make a super powerful transistor but there's no software running on it,

nothing happens, right. Consciousness doesn't just magically occur, So we have to figure out the software side and the code side in order to make this happen. And Lanier argues, and I happen to kind of agree with him that there's no things that we could even do that, So

the hardware is only part of the problem. Meanwhile, you have a lot of people in neuroscience brain experts who roll their eyes at the notion of an artificial intelligence that has superhuman capabilities, because they point out we have a very primitive understanding of our own brains and the concept of intelligence as it applies to human beings, let alone to machines. There are enormous gaps in our knowledge when it comes to how our brains work and how

we process information. We have really good solid foundation in that area, but they're an awful lot of details we just don't understand yet, and there's an enormous void when it comes to the topic of consciousness. We can draw some very broad conclusions about consciousness based off of medical experience. For example, in the medical record, there are plenty of case of people who have experienced various types of trauma to the brain who have also experienced changes in their consciousness.

This reinforces the idea that consciousness arises from the physical matter and the electrochemical processes that are in the brain. But beyond those very broad conclusions, and even these ideas don't have universal adoption. They broadly do in the medical field, but they're plenty of people who disagree with this kind of concept. We just don't know very much about it.

Like consciousness, is one of those things we tend to define by what it isn't rather than what it is, just like intelligence, so many experts in that field say it's premature, to say the least, that we even speculate about creating, at least purposefully, any sort of artificial intelligence,

especially one that possesses consciousness. So even if you do allow for his superhuman intelligence to arrive, making other leaps such as the idea that we're going to find a way to extend our lifespans indefinitely, is a step beyond that, right. It means that you're building an unsubstantiated idea on top of another as yet unsubstantiated idea, which is not the

best way to build an argument typically. Now that does not necessarily mean it won't come to pass in might, but it's not something you can easily support logically, because you're already starting with an assertion that you can't actually back up yet. If this other thing happens, that hasn't happened yet, then this additional thing will happen. That seems like that's a pretty enormous jump in logic. Now, I get the feeling that the singularity was really a topic

of excited debate and discussion about a decade ago. More recently, as we have seen how devilishly hard the problem of AI really is, not to mention how incredibly huge that interdisciplinary field is. Since AI is more nuanced than creating a computer that thinks like a person, that's a very narrow view of AI. I think we've backed off a little bit on the singularity talk. That's not to say there aren't some things associated with the technological singularity that

we should ignore or that we shouldn't tackle. I think it is good to talk about the ethics of artificial intelligence and robotics, for example, including things like whether or not it's a good idea to pursue the development of autonomous weapons, or to allow artificial intelligence to guide economic factors like stock purchases. It's also important to address things

like bias in these artificial intelligence systems. These are all things that are important right now, and they will become even more important if we were ever, to approach the point where we could create a superhuman intelligence, those problems would be magnified a thousand times over if we don't look into them and address them today. It's also important to recognize many of the visions of the singularity come

across to me as somewhat exclusionary. I think a lot of the arguments and the presentations that are all about the technological singularity tend towards the egotistical. Not that the person who expounded the theory is suggesting that he or she is better than anyone else, but rather that somehow that person and perhaps the audience they are addressing, will, for some reason, by default be included in this science

fiction vision of the future. In reality, I suspect if we were to reach any point, especially any point that involves improving quote unquote humanity through biological or technological means, we would see a very exclusive approach to that, meaning you would have a new gap between the halves and the have nots. There would be a disparity so great that it would create instability, or and in some places of the world, arguably everywhere in the world, make instability

way worse than it already is. There are many sciences and technologies that very smart people continue to advance that would presumably be important elements of the technological singularity, and maybe for some of the people in those fields, the idea of contributing toward such a future, striving to get to a technological singularity, that might be part of their drive.

But whether it's a factor or not, many of those areas of study are really important and could end up benefiting us in lots of ways, including many we probably haven't anticipated, even if they don't lead to superhuman intelligence with consciousness us. So, personally, if you haven't figured it out by now, I'm skeptical that any sort of technological singularity is going to happen reasonably soon. I certainly don't

think it will happen within my own lifetime. Now I'm not gonna say it's never gonna happen at all, But if it does happen, I think it's going to be some distant point in the future, and I'm not sure when that will be, but probably long after I'm gone. Also, while I'm at it, I should mention that's actually really

hard to predict the future already. We don't need a technological singularity to make it difficult to see into the future, as you will hear pretty soon when I go back over what my predictions for two thousand eighteen were and we see how incredibly wrong I was. I haven't even listened to that old episode yet, but this is just the way it is every year. And yet, yes, I will be making a predictions episode for twenty nineteen because I never learned my lesson. But if if anything, it'll

be good for a laugh. Right. Well, that wraps up this discussion about the singularity, and while I am skeptical, I also could very well be wrong. That's an important thing to remember. And maybe it will turn out that in twenty nineteen the technological singularity will arrive a little early and I'll be the most shocked out of everybody,

but I doubt it. If you guys have any suggestions for future topics of tech stuff, whether it's a technology, a concept in tech like the singularity, Maybe there's someone you would like me to interview, maybe there's a company you would like me to talk about. Send me an email. The address for the show is tech Stuff at how stuff works dot com, or head on over to our website. That's tech Stuff Podcast. Dot com. You'll find other ways to contact me, you'll find information about the show, and

you'll find a link to our merchandise store. Remember, every purchase you make goes to help the show, and we greatly appreciate it. And I'll to again really soon for more on this and thousands of other topics. Is that how stuff works dot com