What is the Singularity?
The Singularity has nothing to do with the acceleration of technological progress. It is only somewhat related to interdisciplinary convergence. The universe is not specially structured for the Singularity to happen. History has not been particularly leading up to it, except in the sense that inventing new technologies gets easier when civilization has more advanced building tools and knowledge. The Singularity is the creation of smarter-than-human intelligence, nothing less, and nothing more.
The Singularity is not a belief system. It is a likely (but by no means certain) future event with great potential for good and for ill. Sort of like nuclear technology, if nuclear technology could invent more advanced technologies on its own and have independent goals. Kind of scary, really.
The Singularity is a hurdle for the human species to jump, not a stairway to Heaven. It could fairly easily be avoided or delayed, either by blowing up most of the major cities, detonating H-bombs in the upper atmosphere (EMP), someone taking over the world, etc.
The Singularity is not mystical because intelligence is not mystical. The Singularity is just the development of a new type of intelligence. Intelligence operates according to the laws of physics and other rules, just like everything else. It's not magic, though intelligence can sometimes seem like magic when it's greater than our own.
Intelligence is what leads to people like Leonardo da Vinci and Albert Einstein, as well as miracle of human intelligence in general. Remember that every so-called "genius" is still firmly within the bounds of the natural variation of the human species. And our species is more uniform that most. After all, we went through a population bottleneck around 70,000 years ago. Maybe if we were more genetically diverse or went through even more serious challenges as we were evolving (perhaps more vicious, intelligent predators that didn't fall from simple weapons like spears?), then we'd be way smarter than we are now. If that's how history happened, our greater intelligence wouldn't seem "special" -- it'd just be the way things were.
Rather than looking at the Singularity as the culmination of complexity in the universe since the Big Bang, a highly dubious proposition, I look at it as a temporary thing we have to deal with before we can lay back and relax. A single intelligent species on a planet is not a stable state. It's only a matter of time before an intelligent species (like humans) finds out the principles underlying its own intelligence and exploits them to create new variants of itself. In the wider multiverse, this has probably already happened countless times.
Some people say, "you can't engineer intelligence -- it's too mysterious". These are the same people that said life was animated by élan vital, that organic chemicals could not be synthesized from inorganic precursors, that the Earth was the center of the universe, and so on. The Bible or a tendency to pat yourself on the back may have taught you that the principles of intelligence are unbelievably complex or subtle, but that's how most things we don't understand often seem. Mysteriousness is cool, and if intelligence doesn't have mysteriousness, then how can it be cool?
Others say, "human minds are Turing-complete machines, so any other type of mind will have similar capabilities to our own". This is self-congratulatory conceit. Just because two machines are Turing complete does not mean that they can extract statistical regularities from sensory data and arrange them into concepts, inferences and decisions with equal ability. Depending on disparities in the knowledge base and processing structure of the mind, the amount of time it takes to learn something can vary by many orders of magnitude. It appears there are some things certain people just can't learn. Animals can't learn much that humans find simple, even though they obviously have some form of intelligence.
In the same way that someone of average intelligence will never be able to make contributions to the cutting edge of particle physics, we humans will never be able to achieve certain feats with our limited brains. Instead of crying about it or going into a state of denial, we need to come up with a theory of intelligence and use it to boost our own, as well as instantiate intelligence in a nonbiological medium.
It is a mistake to think that the intelligence we create will be on our side automatically, for instance by integrating ourselves sufficiently close with it, or by trusting that wisdom is inextricably connected to intelligence. This is optimistic fantasy. It makes a nice story, but the reality -- that we'll need to work our asses off to ensure that digital intelligence is aligned with our goals -- is far less pleasant. It means we need to reevaluate our conception of the future. The problem -- creating predictably benevolent intelligence -- is absolutely overwhelming once you realize its scope.
Most of the challenges we face as individuals and as organizations have to do with other humans. Convincing them to do things, meeting their expectations, competing with other groups, ensuring structure in our organizations. This problem is totally different. There may be only one chance to get it right. It's not about humans, but about a complex structure that we are just beginning to really understand -- the relationship between cognition and "morality", a shorthand for an extremely complex of human-specific rules and tendencies that many of us mistakenly assume automatically prepackaged with any intelligence.
This should not be a religion or a movement. It's an engineering task. Much more mundane than you might think. The philosophy necessary for success may be complex, but the But just because it's mundane doesn't mean that it won't be difficult or the that benefits of success won't be sublime.
It's a difficult task, but it seems possible. We just need to do it. Even if you're not a programmer or AI theorist, intellectual, moral, and financial support means a lot.
August 9th, 2008 - 07:45
I find myself agreeing, like always – on the definition. A singularity isn’t merely general artificial intelligence smarter than humanity – it is a state of the future becoming intransparant because of that. I can easily imagine a superhuman intelligence emerging and then nothing changing. That wouldn’t be a singularity. The term singularity refers to mathematical states where we can say nothing of what happens next.
Likewise to a “superhuman intelligence singularity” there might be other singularities. With the risk of contaminating the discussion (again) I could label the emergence of timetravel technology holding the potential of a singularity – or the arrival from space of a technological alien species. these are all events with an incomprehensibility component. They are singular.
The problem with so many `transhuman’ terminology (there’s another term!) is that they all deal with unknowns. You cant apply sensible metaphors in any way. People will try to apply familiar paradigms to these fundamental unknowns and those metaphors are all clearly fallible in this case.
August 9th, 2008 - 12:50
All engineering tasks are completed within a context of values, and thereafter reflect those values to some extent. Creation of AGI will not be an exception to this; indeed, if anything, it will be an exaggeration of this. Except accidentally, we cannot create something that does not reflect our values. Thus, the task of AGI creation is not so mundane as you’ve suggested, and should indeed reflect the best of our values — philosophic, religious and otherwise. We may prove unable to avoid accidents or unintentional aspects of creation, but we should try because our future may depend on it.
August 9th, 2008 - 13:06
“The universe is not specially structured for the Singularity to happen.”
Hmmmmmmmmmmmmm…. :-)
If the universe is not specially structured for the Singularity to happen, one might conclude that it will never happen. Since we are in the midst of it now, I would have to disagree. However, greater minds than mine have been wrong in the past.
Hope all is well.
August 9th, 2008 - 15:45
Pigeons are turning-complete machines, so any other type of mind will have similar capabilities to pigeons.
August 9th, 2008 - 20:21
I always wonder about the fact that intelligence is no longer really necessary from an evolutionary perspective. I know its controversial, but I think that humanity’s average IQ is decreasing and this will continue into to do so in the future. See;
http://www.fourmilab.ch/documents/IQ/1950-2050/
Intelligence evolved for specific reasons, a main reason being to avoid death and accidents. Those with a higher general intelligence were more likely to live than those with a lower “g”. Now, though, technology and material wealth has made life considerably easier and thus intelligence is no longer necessary and is not being selected for. If a benevolent socialist artificial intelligence arose tommorrow to create a massive amount of material wealth, wouldn’t that just increase the dysgenic trend among humans? Brain evolution might favor humans being more social or other traits, not necessarily general intelligence. Intelligence from an evolutionary perspective is strictly utilitarian. Why should evolution waste valuable brain real estate on intelligence when it is no longer needed? Even future human evolution through self brain modification would likely follow this rule as well. General intelligence is related to computational capacity of the brain, but it is also an optimization of that capacity(the amount of proteins in the synapses, number of dendrites, how different brain regions interact with one another). Even if you double your brain mass (through brain implants or stem cells) it doesn’t necessarily follow that you would increase general intelligence, especially without some sort of selection pressures in place (like death as in the past). Increasing mass might merely increase the amount of noise but general intelligence would not increase (kind of like a blue whale has a larger brain, but is no more intelligence than a human). The idea of a self improving AI that improves its own intelligence by a huge amount is nonsense. Thats not how evolution works. Evolution always takes the lazy route. That’s my main beef with the idea of the singularity.
August 10th, 2008 - 09:20
Isn’t it self-defeating for someone who has previously written about how “singularity” is a useless term to now be writing about what the singularity is?
August 10th, 2008 - 14:13
I respect Michael and enjoy reading his thoughts, but he sounds sophomoric and shrill when he addresses religion. He may want to pause and take a deep breathe when addressing the subject (his emotional side gets the better of him).
It would be really helpful for him to spend some time with Jesuit brothers or priests and develop a more mature approach to the subject.
August 11th, 2008 - 06:04
“It would be really helpful for him to spend some time with Jesuit brothers or priests and develop a more mature approach to the subject.”
Spare me.
Rather than resorting to a science-fictional God, a Singularitarian would rather direct his efforts towards building a real one.
August 11th, 2008 - 09:46
phil.gs wrote:
Actually, phil; Michael does that all the time. It’s important to him (or so I gather via context) because it allows him to express a greater degree of specificity on the subject — and to operate as a commentary on how the term itself is used. This is semanticism, but it is //productive// semanticism.
Mike wrote:
It’s controversial because it’s not true. The average IQ of the human race is increasing over time; this is called the “Flynn Effect” (google it). Now, if we want to get into discussions about executive function, that’s possibly another story — but there’s really no effective way to measure that with any tests available today.
As to evolutionary necessities; the human race no longer operates under the paradigm of biological evolution. The standard pressures and mechanisms by which evolution operate no longer have any significant impact on our race. (While genetic flow can still occur, especially through plasmid exchange within the germline, reversion to the mean essentially neutralizes this as a factor).
I will make an assertion: “Human evolution has shifted from genetic to memetic mechanisms.“
August 11th, 2008 - 23:17
Phil: if people are going to keep using the term anyway, I might as well push the original, Vingean definition and my interpretation of it.
August 12th, 2008 - 06:45
Michael, Hi, I hope you’re doing well and are open to disagreement. I couln’t disagree with you more about how false your fist statement is. The singularity has a lot to do with technological acceleration.
Second, the singularity appears to have all the qualities of a belief system, although one based on scientific data and theory.
Third I don’t believe the achievement of Artifical Intelligence to be the only result of the singularity. I believe, there’s that word again, transhumanity is possibly a more likely result. Care to debate?
August 12th, 2008 - 09:26
The Singularity – by my current best guess – will occur as a result of some self improving software process, and this will be before the 2020.
I guess, before the year 2013 it will be possible to launch such a self optimizer and that it will be quite harmless.
The question is however, how dangerous will be the team who will develop it to the initial phase.
But that’s exactly, what’s gonna happen, like it or not.
– Thomas
August 13th, 2008 - 12:42
Belief in “the Singularity” is clearly becoming the 21st century religion of choice for the uber-nerd. This isn’t necessarily a bad thing, but we should be wary of any movement that seeks to “improve” humanity. Perhaps a bit more grounding in history, philosophy and the arts would serve this community well; there is a long history of absolutist, utopian, hubristic intellectual movements wreaking utter havoc on the rest of humanity.
For some slightly outside the box thoughts on the Singularity, Transhumanism and the like, please see:
http://singulatarian.blogspot.com/
August 13th, 2008 - 13:36
You spelled Singularitarian wrong.
August 20th, 2008 - 15:59
Good thing the cavemen among us, the luddites, and the supernaturalists never win. They’re infested with vestigial memes with – unlike technology – absolutely no purpose, no power to affect the world, except negatively – “don’t research this, because [favorite fictional character] says so”, and killing for their memes. As a technologist, I see them merely as wasted, useless brains, in need of help.
I for one welcome our new AGI overlord(s).
August 26th, 2008 - 11:16
Michael,
Interesting article. From different frames of reference:
As a Christian, the “Singularity” may be the Anti-Christ; ‘…Man of Men…’. Not necessarily a good thing to strive for, yet a possibily enevitable conclusion (or checkpoint).
As a scientist, the Singularity is the ultimate carrot to which we reach. It is perhaps the ‘what?’ of the great ‘why?’
As an economist, a Singularity is merely a temporary solution that will create a variation in the present system. The new system will become obsolete and the neccesity for a new “Singularity” to evolve and the cycle continues. In this sense, every seemingly original technology is indeed a singularity, even if only for a little while.
- Josh