The idea that there is coming “technological singularity” is a somewhat popular one these days. Specifically it is claimed that technology will continue to advance at an ever-increasing rate until we reach a point in the not-to-distant future (usually within 100 years of the time that the prediction is being made) where technology will be advancing so fast that it is able to give us everything we could possibly desire and more. Among the promised rewards of this singularity are immortality and the ability of intelligence to spread throughout the universe at a speed close to that of light. To me these predictions remind me of those made in days past of flying cars and regular trips to the moon, just with different fantasies. There are often rapid bursts of progress within some specific domain, and extrapolating from those bursts of progress will seem to imply that soon technology will be doing all sorts of miraculous things. But so far that is not how things have worked, the burst has tapered off and rapid development shifts to a different domain. The cars we drive now, for example, aren’t that much better than those that were around twenty or thirty years ago; they may be more comfortable but they don’t go too much faster, nor are they that much more efficient (some are even less efficient). Currently the burst of innovation is in computing technology, so naturally the cool future things are expected to be AIs and such rather than flying cars. However, I already see indications that the rapid advancements made in computers are already beginning to taper off; for example, the word processors we have now aren’t much more powerful than those of five or ten years ago, they are just more comfortable to use. Just because more powerful computers will continue to be developed doesn’t meant that the power will necessarily be put towards productive uses. And if I was to mock the position further I would point out that it plays on people’s desires; they want to live forever and to have all their desires satisfied. Thus people are likely to believe in the singularity simply because they want it to be true, which interferes with their ability to rationally evaluate its likelihood. And, finally, I will note that some of the things that are claimed are highly unlikely to happen no matter what technological capabilities we develop. For example, it is highly unlikely that we will ever expand into the universe turning it into “smart matter” as we go, as one prediction I read claimed. Any highly advanced civilization is going to be using as much power as it has available to it, which means reducing “leaks” into the external universe as much as possible, both through stray transmissions and through purposeful egress. From the point of view of the people remaining behind any energy spent on spreading out to the rest of universe would be wasted. This means that either everyone will stay home, or everyone will leave, creating a society that acts like locusts, exhausting the resources of one region of space and then moving on to the next, or they will set things up so that the energy spend expanding outwards is returned many times over by energy sucked out of the rest of the universe. For the sake of the rest of the universe we had better hope that the first is the preferred option.
But of course all of that is a digression, because whether there really will be a technological singularity in the future isn’t affected by when it will occur, the similarity of the prediction to flawed predictions made in the past, or whether many of the absurd claims associated with it may turn out to be false. To see whether we are destined for a technological singularity or not we have to examine why this singularity is expected to occur, not simply extrapolate from current trends. And to do that we have to do more than those who actually make the predictions. So is progress really accelerating? And if so, why? To begin with we can start at the level of individuals, because it is individuals who make the discoveries that count as progress. Thus there are two reasons that the rate of progress might accelerate, either individuals are becoming faster at making discoveries, on average, or the number of individuals is increasing. And certainly the population of the world is increasing, and during the periods in the past where technology has been increasing rapidly indeed there were often rapid increases in population as well. And so we might expect, from that observation, that the rate of technological advancement will level off as the number of people levels off. Still, while that can account for some of the rapid advancements we have made in recent times I am reluctant to pin down all of our progress to simple increases in population. First of all the people responsible for new discoveries are only those living outside of the third world, and the third world is where population has been increasing most rapidly. And so the pace of progress doesn’t seem to quite match up with population increases. For example, the population of the United States has doubled within the last 60 years, but the rate of progress may legitimately seem more than double what it was in the 50s. Thus any increase in the progress beyond that must be attributable to an increase in the rate at which individuals can make discoveries.
But what would make someone able to make discoveries faster? Well a better education might, but I highly doubt that we can attribute such progress to better schools. It must then have something to do with the progress of discovery itself, which comes in three parts. The first part of discovery is the idea, the idea is then refined into a precise form, and then the precise form is tested. And all of this is a relatively linear process, there are no shortcuts, and, while someone might try to work on more than one idea at a time, that doesn’t make them any faster at making discoveries, since the time taken to work on one idea detracts from the time devoted to another. We might legitimately claim that technology has made the second and third steps much faster than they were before, that computer models and mathematical assistance make turning ideas into precise forms and testing them easier than ever before. But, while this could be responsible for the observed increasing rate of progress, it clearly can’t give rise to a singularity, because even if the time for the second and third steps was reduced to zero by some miracle of technology there would still be a relatively fixed amount of time required to develop new ideas. Which means that we would end up with perhaps a very fast rate of progress, but there will be some maximum the rate will increase to at which point it will go no faster.
Naturally this is complicated by the fact that discoveries are not made independently of each other, for progress to exist the people making the discoveries must have been exposed to previous advancements and must be building on them. This might seem like another area in which technological advancements might speed up progress, but communication never seems to have been a major issue in this regard. What is a problem is that it takes time for individuals to absorb and understand new discoveries so that they are able to build on top of them, which is not something that can be sped up by technological advancements. And this time can become a serious limitation when discoveries are made rapidly, because then individuals are forced to decide whether they will spend all their time trying to master the work of other people, or whether they will ignore some of it in order to do original work. No matter what balance is struck there are disadvantages: the more time they spend learning about other discoveries the slower they make their own discoveries, but the more they don’t read the more likely it is that their discoveries will essentially repeat work already done, which also effectively reduces the rate of new discoveries. Of course clever technology can mitigate some of these problems by helping people make better decisions about what to pay attention to and what to ignore. But it is one more hidden limitation that serves to limit how fast progress can be made.
Some may point out that these supposed limitations are all human limitations, and indeed if we were smarter maybe we could come up with new ideas more often, or ones that were more likely to pan out, and maybe we could assimilate new knowledge quicker. And so some believe that the singularity will occur because of artificial intelligences who will be able to exceed us when it comes to those things. And I agree that artificial intelligences will eventually be developed and that eventually they will be intellectually superior to us, simply because they will be able to think faster, if nothing else. The problem is, however, that artificial intelligences don’t seem like they will be developed anytime soon; even though people have been claiming that they are right around the corner ever since the idea was thought up. Despite all the work that has gone into the field we have yet to develop a machine that can learn as well as a pigeon. But when artificial intelligences are developed I doubt that they will produce a technological singularity. Although they may be superior thinkers to us there are still upper bounds on the speed of computation, which means that there will still be fixed limits on how fast they can make discoveries. Furthermore there is the problem that discovery works by trying our new approaches to problems, new ways of thinking about things. As more and more discoveries are made it will probably become harder and harder to find productive “new” ways of looking at things to make further advancements, as all the obvious approaches will have already been tried, forcing investigators to consider more and more bizarre possibilities, which are much less likely to prove fruitful. Which is an argument that sometime in the distant future after the maximum speed of computation and maximum populations have been reached that the rate of progress will actually slow down.
So, as nice as a technological singularity would be, I highly doubt that one will ever be encountered. Although the rate of progress may increase further, and indeed it would be nice if it did, it will eventually run into some upper bound. And we will certainly never achieve the kind of power where everything is possible. Of course some may think that I am simply ignoring the possibility of discoveries that turn everything we know on its head and remove those limits. First of all I would point out that neither can we rationally count on such discoveries since we don’t even know what they are. And, secondly, claims that something is impossible made nowadays are much more credible than those made in the past and subsequently overturned, for the simple reason that in the past the claim that something couldn’t be done was based simply on an inability to envision how it could be done. But now when we say something can’t be done it is because we have uncovered what look to be physical laws including such constraints. And although we have been mistaken about physical law before we have never been as drastically mistaken as we would have to be if these limitations were to disappear.