This, gentle reader, is how you are likely to feel at the end of today’s entry.
I was browsing the latest issue of The Economist, where they commented on the plans for a new high-speed railroad in the UK. They did not much like it, thinking it was bad economy. But what struck me was reading that a certain stretch of railroad was supposed to be finished in 2043, and I thought: “Humans will use railroads in 2043??” and then, “Will there even be humans as we know them in 2043?”
I personally assume that there will indeed be humans similar to us in 2043, unless some global disaster befalls us (like genetically engineered viruses, man-made black holes, or the unexpected wayward asteroid). Looking up, it seems that Ray Kurzweil agrees, the Singularity is not until 2045. Close call though. I am pretty sure it was 2040 a while ago, and even earlier some years ago. I guess programming artificial intelligence is harder than expected, especially when Kurzweil is no longer doing it himself. (He is the guy behind the product that eventually became Dragon NaturallySpeaking, the speech recognition software that is better than a professional human at transcribing speech. Some training required.)
The Singularity, strictly speaking, is when artificial intelligence (or artificially augmented human intelligence) starts a runaway process of self-improvement at an exponential rate. Whether this ever happens is still hotly debated. But there is something else that is not: The knowledge explosion.
You may have heard of Moore’s Law, about the progress of information technology. It has taken on its own life — in its older form it said that the density of integrated circuits doubled every two years, now it is widely cited as “the performance of computer hardware doubles every 18 months”, which is roughly what we have experienced for the last couple decades. But others have expanded this law into the past (the rate of technological progress has increased for thousands of years, although in fits and starts) and even into biological evolution, which also seems to have happened faster and faster: Life remained single-celled for some 3 billion years if not more before it started getting more complex.
Anyway, the sum of all this: The accelerating acceleration of change is accelerating. Change is not just happening faster and faster, the pace at which it happens faster and faster is itself getting faster. The knowledge we have allows us to create new tools that let us gather more knowledge faster, and this new knowledge lets us gather more knowledge faster again. We are quietly approaching the point (if we aren’t there already) where the sum of gathered data at the beginning of the year is less than what is discovered at the end of the year. Of course, most of these data are just more details about things we already know in outline, but it is still a pretty mind-boggling thought.
At some point in the lifetime of today’s middle-aged people, this process is expected to reach the level where knowledge doubles each DAY. Ah, thank you Google: “Currently, Kurtzweil estimates that knowledge doubles every 5 years. By the year 2040 it will be doubling every day.” If life expectation continues to increase by 5 hours a day as it has done lately, a good number of my classmates should be around to see it. I would not mind being there too, but of course it is no certainty.
Now, give this a brief thought. What you do today, the way your life is heading into the future – how relevant will that be in 2040 – plus minus a couple years – when human knowledge doubles Every Day? Mind you, that is the human knowledge of 2040 that doubles every day, which is in itself thousands of times the knowledge we have today. Because it has doubled every week, month and year for some time. In fact, if it miraculously only doubled every year for these 29 years, it would still be 2^29= 536 870 912 times more than now.
In reality, this speed is expected to be surpassed in a few years. And this wealth of information, hundreds of millions of times more than today (more realistically billions of times more) is what will double every day, or perhaps it is down to every hour or two in 2043, at the time when this railroad is supposed to be finished.
Will we still take trains from London to Leeds when we know a billion times more than we do today? And will learn a billion times more than we know today over the duration of the train trip?