How long til the Singularity?
My guess for the time until we learn to create artificial consciousness to specification is on the order of half a million years, maybe never.
Sorry folks, but in order to devise an artificial mind to specification, humanity would have to have solved not one, but several foundational mysteries of the universe such as the mind-body problem, nature v. nurture, the relationship of symbol to reality and thought to symbol, not to mention Godelian paradoxes relating to reducing self-awareness to a finite logical system.
We are not talking about something like heavier-than-air flight, where the general principles are understood, natural examples of birds hover before our eyes, and the only question is designing a tool to do something we know can be done.
Instead we are talking about questions which have baffled mankind since the dawn of history, on which we have made precious little to no progress, and which are not open to empirical investigation. There is no way to measure consciousness or analyze it, much less self-aware consciousness. Armed only with a yardstick and stopwatch, how does one convince a solipsist that consciousness aside from his own exists?
Does anyone really think the nature of Thought, or the hidden causes of Inspiration, Art, Morality, Logic, Reason, Emotion, Passion, Memory, Intuition, and the relationship between Brain and Idea are going to be discovered any time soon? We do not even have a scientific definition for the minimum unit of thought processes.
What is the relationship between the firing of a synapse and the minimum syllable of a concept that conveys meaning? Anyone? Bueller? Anyone?
By “to specification” I mean that a self-aware mind can be programmed and reprogrammed according to a scientific understanding of the causes of thought. Your computer is not in love with Isolt; you enter the proper parameters in the proper way; now it is as infatuated as Tristram. Your computer is a Bimetallist; you enter the proper codes to influence its subconscious judgment process that it uses when thinking about political economics, you reshape its sense of right and wrong, you rebalance the weight it gives to evidence, and rewrite its gut-instinct view of the universe, and lo and behold, now it is a Keynesian.
Because if you cannot write a program to be a self-aware Bimetallist, or Buddhist, or Basketball fan, then you cannot make a positronic brain to obey the Three Laws of Robotics. I do not see an innate difference between designing a brain so that “Thou shalt not harm a human being, or through inaction, allow a human being to come to harm” and “Thou shalt prefer Impressionists to Pre-Raphaelites.”
If your science has reduced consciousness to a series of mechanical operations, setting up the code to carry out the operations of Asimov’s Three Laws cannot be any more or less difficult than carrying out the operation of the Four Noble Truths. There may be specific technical difficulties making some operations be carried out, but there is nothing innate in the concept itself: if you can program a mind to value, at a high priority, some extremely abstract concept like “Obedience” or “Cooperation” or “Nonviolence” then it is a safe assumption that you can program a mind to value at a high priority some semi-abstract concept like “Basketball” or “Detachment” or “Economic systems using two hard metals as currency.”
The distance between our current theory and practice of psychology, and the engineering needed to design minds to specification, can be grasped by contemplating the gap in our understanding. What makes a man infatuated? Why does he join one political party or denomination as opposed to another? Why are some men dreamers and other men practical? I do not mean that we do not have airy speculations about the hidden roots of the human mind. I mean only that they are no more scientific, no more based on firm empirical data, than Medieval speculation that melancholia was produced by melancholic humors in the blood.
Could science design artificial minds to specification without any specifications?
You can make the argument from ‘game theory’ that there are certain innate strategies of cooperation and competition to which any mind should eventually evolve. This is a metaphysical assumption that there is a moral algorithm built into the universe itself, which any sufficiently rational mind will discover. If this metaphysical assumption is correct (I have never seen an argument proving the assumption), then no matter what the starting point of the artificial brain, it will eventually reason itself to the conclusion congruent with whatever the universe has built into it as correct for the proper weight to be given cooperative and competitive strategies.
But if your science is in the business of making minds, then “sufficiently rational” is something you have reduced to a measurable magnitude: brain-atoms in configuration such-and-such are reasonable people, and brain-magnitude so-and-so makes people geniuses, but brain atoms in configuration this-and-that are neurotic, or violent, or proud, or lustful.
I note that the human race is not sufficiently rational to reach agreement on this point, or even on the point of whether there is an objective morality that reason can discover for us. Granting that, in order to posit an ability to design artificial minds to specification without any specifications, we would have to posit a science of artificial intelligence that can create a perfectly rational superhuman intelligence, but not be able to create a merely human semi-rational intelligence. In other words, we would have to assume the universe just so happens to be made, so that apish minds can produce angelic minds with minimum effort. Naturally, while it might be nice if this were the case, the opposite seems much more likely. What happens when an artificial intelligence is damaged, or suffers a bug?
But suppose we are talking about creating an artificial mind not to specification. Now, if we are talking about merely creating the conditions under which consciousness arises, setting up an environment where a simple set of reflexes can evolve into a self-aware being, well … every Mom who has raised a child knows how to do that, and Moms have been doing this since the Human Race rolled off the assembly line back in the prepaleolithic.
About John C Wright
John C. Wright is a practicing philosopher, a retired attorney, newspaperman, and newspaper editor, and a published author of science fiction. Once a Houyhnhnm, he was expelled from the august ranks of purely rational beings when he fell in love; but retains an honorary title.
