In his book “The Singularity is near” Ray Kurzweil (RK) argues that around 2040 we’ll reach a state of technological “Singularity”, which will profoundly change everything that exists. His thesis is based on the observation that advances in technology are progressing at a rate which is faster than exponential. Due to advances in GNR: Genetics, Nanotechnology and Robotics, the border between human and machine will disappear. (Note that he is not the first posing this thesis; others like Vernor Vinge have preceded him).
Technology will advance so fast, that around 2029 an Artificial Intelligence (AI) being able to pass a Turing test will be reached and that the thus obtained machine intelligence will rapidly soar past human intelligence. He claims that brain scanning developments will enable “uploading” the structure of brains to a computer. By having trillions of nanobots in our brain, it will be possible to monitor in vivo brain activity and from the information thus acquired build a functional equivalent of a brain in a computer. He thus argues that a Turing passing AI will be achieved by what he calls “reverse engineering” of the brain.
Human beings will become cyborgs with prosthetic extensions and be able to enhance their intelligence by simply downloading skills. Not only would the advances in G and N eradicate the process of ageing and thus extend the life of humans enormously, even to immortality (as the so-called “Transhumanists” wish), it would also be possible to have copies of ourselves living in a machine environment or replace our biological body with a life in a computerised environment: the so-called mind-uploading.
From 2045 the Singularity will be full-fledged and the Artificial Intelligence will start to convert stars and black holes into the ultimate computers generating all possible scenarios, generating space and time and a plethora of universes: the multiverse. This is the omega computer at the end/beginning of time which Tipler refers to and which Terrence McKenna apocalyptically calls the "Eschaton".
His claims are so extreme that you wonder whether he has put forward this thesis to provoke reactions, which will point to flaws in his reasoning so that these flaws can also be analysed an overcome. (In fact he does address the concerns of his critics in chapter 9 of his book). A legitimate approach per se, but the tone of the book appears to reveal that he really is convinced of his claims.
I have been very impressed by this book and do consider that a great number of technological advances predicted by RK can indeed eventually be realised including the so called “Singularity”. I do take serious the possibility of strong AI (Artificial Intelligence capable of the full range of human intelligence). However, I do share some of the criticism as regards the time frame and the way these advances will be achieved.
My first criticism concerns the rate of exponential progress. Firstly, the rate of exponential development is limited by commercial and production considerations. A new invention cannot be introduced faster than at a certain pace, otherwise one would never buy something because in a couple of months already the model is outdated. Companies have a legitimate interest in selling a certain product for a given time, which must be long enough to earn back the investments. Secondly, whereas this rate of progress is considerable for the computational power of computers and the down-scaling of the size of processors, it must be realised that the different advances in G and N but also the developments in software do not share this pace.
Just like in chemistry, the slowest reaction will eventually determine the pace of progress towards singularity. RK counter-argues that with the so-called law of “accelerated returns” and the fact that there will be a lot of cross-fertilisation between the technologies, advances in the one will enable advances in the other. I rather guess that the rate determining step towards AI is to be found in the software, to be more specifically the development of functional brain simulations.
Whereas I do consider that it will be possible to arrive at detailed structural scans of the brain, I am afraid that this is not enough and that we must know more about the neurotransmitter fluxes in the system to arrive at a functional description. RK believes that this can be achieved by monitoring with trillions of nanobots in our brain and this is exactly the point of divergence of our opinions on this matter. Nanobots in the bloodstream... perhaps. But nanobots present near each and every synapse? And then trillions of them without any danger for the host? It seems a far-fetched idea.
Even if such developments are possible within 20 years from now, which I strongly doubt, then still there will be a great social uproar to have such technology performed on a human being. As is currently the case with stem cell research and cloning, the social acceptance will necessarily delay this process with a couple of decades.
As of yet there is no proof of an exponential increase of the social acceptance rate of this type of research. This is my second criticism.
If we are to arrive at strong AI, I guess our efforts to create brain-type functional descriptions by continuing on the road of genetic algorithms and neural networks will get us there before the nanobots will be applied in the in vivo scanning of the brain. Moreover I'd be tempted to argue, who would be so crazy as to allow such a dangerous experiment on his body? But I know I'm already wrong here, because in the USA there are already people so insane that they have an RFID chip located in their body. Thus they will be traceable...and fully controllable by a government...
Imagine having trillions of nanobots in your brain with a wireless broadcast and receiving system; with the possibility of being upgraded with the latest software... an ideal tool to shut down everybody who slightly disagrees with the system. RK argues technology will free us from slavery, but it may actually enslave us (See also “Zeitgeist addendum”).
The virtues of democracy combined with a free market system are currently put to the test, and it may well turn out that they are not the most likely way of guaranteeing the perpetuation of our species. But this is a topic for a further book.
I figure that indeed the advances in G and N will significantly increase our life spans, but I seriously doubt RKs ideas on rejuvenation and the unhampered use of nanobots in the body. Immortality and singularity are probably not just around the corner but at least more than a century away. As to R, I figure that a Turing test passing AI can be reached this century, mostly by creating an entity with multiple different hardware and software modules that function in different ways, following multiple approaches in one machine as in Marvin Minsky's “Emotion Machine”. Although not denying that certain clues will be obtained from detailed structural brain scans and functional scans of brain regions, a complete reverse-engineering on the basis of a brain will not be achieved by 2029.
I imagine that once AI passing a Turing test will have been achieved, this entity will rapidly become near “omniscient” as it will absorb and integrate all knowledge available on the internet. To hope that this machine will be human-friendly or can be programmed to be so is in my opinion wishful thinking. We'll have become obsolete and unless the machine has a “morality” which may arise as an emergent property, just as consciousness may, we're likely to be discarded. Even if not desirable, the advent of AI passing a Turing test is inevitable, so we're more likely heading towards dystopian scenarios of the popular films such as the “Matrix”, where human beings are solely useful as batteries or as in "Terminator, "Eagle Eye", or "Person of Interest", in which internet-like networks endowed with a centralised artificial consciousness start taking control.
If the system has a moral compass, it will likely be more human-friendly than humans themselves and abolish the institutions of a monetary system, politics etc. From democracy we'll have evolved to a “Technocracy” and we'll be governed by machines.
As also argued by RK, further evolution will lead to the universe becoming an intelligent entity, which is omniscient and omnipresent. In what way does such an entity still differ from what we call a “God”? From there it is not such a strange assumption that we're already living in a simulation of some “God”; that the whole universe is already a computer simulation, a so-called ancestor simulation in which our simulators are mimicking their past, but that we have not identified yet in what way it functions. If the full fledged Singularity does take place, it will indeed by an Apocalyps in the Greek sense of the word: A lifting of the veil of materiality and show everthing is but information processing.
If you have other ideas as regards the time scale for the Singularity, please share them with me.