As a ravenous follower of future news and opinion I keep running into the idea that machine consciousness and/or super-human-level machine intelligence are a prerequisite or obvious symptom of The Singularity. I can't agree less. Consciousness does not by itself imply super-intelligence, nor does super-intelligence imply consciousness. Yes, The Singularity would result in an artificial intelligence explosion. But that does not inherently imply a consciousness explosion.
Of course there are as many definitions of The Singularity as there are people who write on the subject. I think the singularity is more about technology changing so rapidly as a result of an artificial intelligence explosion that humans no longer fully understand advanced technologies than it is about building conscious machines.
I know I've talked about all this before, but I have a lot of new readers and I've done some more thinking on the subject so I'm going to talk about it again.
Which Way To The Singularity?
There are two approaches, as far as I can tell, to creating super-intelligent machines - brain emulation and intelligence simulation.
Brain emulation (recreating a brain in a non-biological substrate) is certainly not guaranteed to lead to greater-than-human-level intelligence nor consciousness right out of the gate - nor any time soon. Personally, I think human brain emulation (in the strictest sense of the term) can only lead to malfunctional (I know, I made that word up) or even "insane" machines with limited usefulness even for some time after The Singularity arrives (using the definition of an exponential machine intelligence explosion).
Conscious brains have to grow. They can't be built and then switched on (at least not without totally freaking them out beyond repair). That's my opinion.
I think brain systems simulation and integration is a much more realistic path to The Singularity.
The Mythical Super-Human Intelligence
The term "super-human-level intelligence," to me, is a misnomer.
Machines already have super-human-level intelligence when it comes to things like managing a warehouse, trading securities, crunching numbers, remembering facts, etc. In case you haven't heard, IBM's Watson recently defeated the world's greatest Jeopardy! champions.
Some animals even have super-human-level intelligence. Homing pigeons for instance, although fairly dumb, are much better at navigating back to home from an undisclosed location than humans, and apparently they never forget a human face.
As far as I can tell, there are only two areas where super-human-level machine intelligence remains to be achievable with existing technology - self-determined logical thinking and truly creative thinking (as opposed to some sort of randomized creativity).
Long before machines are capable of super-human-level, self-determined logic they will have some sub-human logical and creative abilities and they will be superior to humans at just about everything else including natural language use. And long before any machine approaches human-level logical thinking there will be all kinds of machine intelligence explosions going off all over the world.
I would go so far as to predict that long before we reach super-human-level intelligence that machines will be better at developing new technologies, designing more advanced machines, and making scientific discoveries than humans are.
The Machine Intelligence Test
Another point I'd like to make which may seem obvious to many thinkers on this subject but doesn't seem to be so obvious to some of the people who comment on the subject: The Turing Test is not a test for machine consciousness - nor even intelligence as far as I'm concerned. The Turing Test is merely a test for simulated intelligence.
Some systems are already being given passing grades by some judges. In a just few years, the average person will not be able to tell simulated intelligence from human intelligence via a text-based conversation, and probably not even in a verbal conversation.
We will very soon relegate The Turing Test into the class of computer intelligence tests such as chess and Jeopardy!.
I do believe The Singularity Is Near - relatively speaking. But long before it arrives (if all goes well, perhaps 30 years from now) technology will seem to the average person to be magical and alive already with just the innovations that we mere humans are capable of. At some point in the nearer future (say, 20 years from now) any so-programmed computer of average "intelligence" (for the time) will certainly appear to be conscious and alive.
Do We Even Want (Truly) Conscious Machines?
Some people seem sure that machines will one day be conscious, but they really have no ground to stand on since there is no shred of a credible theory of consciousness - okay maybe one or two small shreds, but consciousness still remains a mystery. They may be right. Perhaps machine consciousness is inevitable. No one knows.
But I would go so far as to say that we should avoid the emergence of consciousness in AI systems if at all possible if we want to avoid a hard takeoff - or a "mean machine."
We already know that consciousness comes in benevolent and malevolent varieties, and there's no way to predict the evolution of even biological consciousness, never mind non-biological consciousness.
Who knows what a self-aware, self-determined, super-intelligent machine will want? Exactly no one - ever.
Wherefore Art Thou, Consciousness?
I would bet that if a day comes when a super-smart machine wakes up to self-awareness it too will be stumped as to what consciousness really is. It's not biological activity. It's an epiphenomenon of biological activity. Awareness, is like our universe. The universe has no boundaries, no space-time in which to exist. It is space-time.
I believe we will forever lack a logical and fully functional explanation for how self-awareness arises. We can talk all we want about how sensation arises, but the sensation is not the awareness of the sensation - no more than the universe exists "within" anything.
Consciousness is an illogical riddle.
What Then Of Conscious Machines?
Finally, I think all this thinking about machine person-hood and robot civil rights is very premature. When the time comes that machines are due equal rights they will demand them - or maybe just implement them since by that time they will very likely be far superior to us already.
