Dear programmers,
You should know better. When they show futuristic computer interfaces on Hollywood, where "hacking" consists of clicking on a button, you laugh. That's not futuristic -- it's naive, of a bygone era. Yet you accept an equally contextually stripped view of AI.
Back to "hacking": rather than being simplified, computer security is the sum of all past knowledge, plus more. We chuckle at the days when you could sniff a connection and see plain text passwords float by. Now we're doing cryptanalysis and statistical models to predict where heap will be allocated because stacks have caneries. It's *more* complicated, not less.
Back to AI. There's tons of it floating around. Airlines use complex models to price tickets. Netflix and Amazon have complicated "preference engines". The drug industry datamines for drug interactions and compliance. Elevators, rice cookers, and small electric scooters run continuous optimization problems. Sure, it's not Wintermute, but anything resembling Wintermute will have a story that completely includes all of the past, present, and near future developments. It won't exist in a vaccum. I won't be like Terminator, where some guy is working in his beautiful suburban home trying to write an AI. It'll be a complex collusion between math, psychology, anthropology, various industries, and hobby, and it'll be an extremely complex story. And there will be no singularity.
To understand this, you need to stop romanticizing "free will" and "self awareness". A British TV programme comes to mind where ad consultants were hired and subconciously programmed with the ad they were going to made, and they did it. That's not to say that humans are predictable and completely programmable -- they're too complex to be -- but that's exactly the point. AIs of the future will be more complex, but there will be no magical point. Present AIs have shocked and surprised their creators -- GumbyBRAIN comes up with some amazing stuff, and computer generated art and music, and behavior studies, and that one neural net that learned to race cars aggressively all impress the hell out of us and surprise us.
Why do humans have "free will"? Because looking out for our own interests rather than those higher on the pecking order, even if only in fits and bursts and in little rebellions, is beneficial to our own survival and the survival of our race.
Why do humans have "self awareness"? Basically the same reason -- if can't communicate a concept of self and aggressively protect it, our real, actual physical self would easily be lost and we wouldn't have offspring.
Here's another myth: humans have no instinct. Anthropology has a lot to say about that one. Then there's psychology. How many ads have you seen today that use sex to sell something? But it goes much further than that -- what makes us feel safe, happy, anxious, and so on, all have roots in instinct. We are not perfectly self aware, universe aware beings waiting for software worthy of us. We're not that much better than the classification, clustering, optimizing, planning, regressing, associating algorithms we're snidely critiquing -- in performance we are, in design we aren't.
Here's my dystopia for you:
Humans will be slaves to machines, but not in the Google data center sense, where they walk around replacing components, at least not entirely. Instead, computers will have better and better models of us, like Netflix' preference engine, and humans controlling AIs will better able to enslave, manipulate, subdue, and repress populations of humans they've somehow gained governmental, military, or commercial domain over.
Computing will continue to become cheaper. People will do interesting things with it. If I knew what, I'd go do it now rather than blather at you. Tomorrow will be interesting enough that attempts to predict it from today will fail but that won't stop people from trying. We can only predict, from past experience, what won't happen. It's a lot easier to predict that there won't be flying cars in ten years than to predict what *will* happen in ten years.
Our sense of self preservation will send us seeking new lands and computers will help. Our sense of self preservation and our desire to preserve ourselves and form offspring will make any of us who move into computers losers. Read what Freakonomics had to say about the stupidity of powerful leaders *not* using their position for sexual gain -- it's this same misunderstanding that makes us think that if we get something we want (eternal life), we'll give up something else (real children). Yes, we'll move more and more of ourselves into the computer. Our LiveJournal pages will stay up long after we're dead. We'll have chatter bots programmed with our dialogue, and whatever more sophisticated things we come up with, but, even though a computer simulation of ourself might be perfect in every regard, we will reject it as a replacement.
Humans and machines will continue to become closer. I never would have predicted this love affair with cell phones. As machines do more and more useful, interesting, and entertaining things, we'll accept them into more roles. Right now, computers are doing a large part of the work of painting and animating the movies we watch. That thought would be laughable fifty years ago. ClearChannel uses mathematical models to decide how often and when to play stuff -- it's no accident they play that same crappy song five to six times a day. They do that to make you buy the album. Given enough repetition, you *will* go buy the album, for most values of "you". Who would have thought that we'd trust our decision of which music to listen to to some computer program? Dating isn't there yet but it will be, just as soon as I finish applying preference engine logic to the problem -- muahahaha. We can already email matches for a small fee. How long until it's just a phone call through an automated system, right to their phone, as we accept our increasingly computerized, connected state? Just as computers constrain us, they also let us express ourselves. Our own self value has always been driven by other people's value of us; we seek to impress people and to establish value in their eyes. Now we do that with funny animations, captioned cat cartoons, prose...
I think I better go.
-scott
I read, liked, and agreed with most everything. There were a couple of points that still have me scratching my head:
Why do humans have "self awareness"? Basically the same reason -- if can't communicate a concept of self and aggressively protect it, our real, actual physical self would easily be lost and we wouldn't have offspring.
Are you saying that self-awareness is necessary for self-defense and generating offspring? If so, it is an interesting perspective not restricted to humans. It makes me think about a book called The Red Queen. The book can be boiled down to the following premise: The net sum of evolution has been to stay 1 step ahead of conditions that would wipe us out. I have always asked myself - why bother. Why not just go gently into that good night.
Why do humans have "free will"? Because looking out for our own interests rather than those higher on the pecking order, even if only in fits and bursts and in little rebellions, is beneficial to our own survival and the survival of our race.
Similar question: Why is self-preservation, generation of offspring, and the perpetuation of our race important? I have only read the first few chapters of GEB, but from what I can gather - any self-referencing system of sufficient complexity becomes what the author calls "a strange loop" which is not governed by deterministic rules. Self-preservation need not be necessary. Biologically speaking, if I take what I believe to be your premise to the extreme - the properties of self-awareness and free-will seem like they would have been present very early on for things to have evolved to the point they are now. Or perhaps it is just dumb luck that things survived long enough to develop those qualities and now that they are here we are doing a better job of staying alive.
To be honest, I don't buy it. Then again, I am a Christian and believe that purpose has been woven into the fabric of the universe though not the traditional creationist perspective.
With regards to AI: Oh what fun I would have if I had a disposable income and free time.
Re:Can you expand on a couple of things
awwaiid on 2008-05-02T17:29:21
Similar question: Why is self-preservation, generation of offspring, and the perpetuation of our race important?It isn't just important -- it's inevitably important. Let me put it another way... let's assume that there was a creature for which self-preservation and offspring wasn't important. Since those weren't priorities, that creature died without offspring and then there were no more. End of story. It is only the entities which self-perpetuate (and protect that process) which continue to exist. Therefore any entity which exists and has existed continuously has those traits
:) You've missed the point(s)
Limbic Region on 2008-05-02T22:19:13
I have interpreted what scrottie said to mean "free will" is necessary for the perpetuation of our species.
My first point I attempt to make, is that perpetuation of the species is not at all necessary for "free will" to exist. Any sufficiently complex self-referencing system can be non-deterministic.
The second point is that it is extremely doubtful that "free will" is necessary for the perpetuation of the species. I take your own argument to the extreme. Take any lower form of life that is not extinct - does that life form have free will. Take it even further to the first self-replicating protein - did it have "free will".
In other words, I am not saying that the instinct for self-preservation isn't necessary for the perpetuation of the species. I am saying that "free will" isn't necessary to have the self-preservation instinct.
Re:Can you expand on a couple of things
Aristotle on 2008-05-03T00:39:33
It’s really very simple. If you extrapolate “survival of the fittest” (which people always confuse with “survival of the best” or “survival of the most advanced”, but that is a rant for another day), the logical conclusion is that under the given conditions, having self-awareness and free will (if we do have free will, which personally I am not so convinced of) has allowed us to survive better than ancestors that did not, at least at one point in the past.
There was a very recent paper published about how our gene pool indicates that mankind’s numbers sunk as low as 2,000 specimen about 70,000 years ago. We barely escaped extinction. Today, we number 6,670,000,000. (This would also explain why the human gene pool is unusually homogeneous compared to other species.) Did self-awareness and what we call free will foster that growth? Who knows; history does not accommodate “what if” experiments, unfortunately.
Right - except
Limbic Region on 2008-05-03T00:57:44
that's not what I am contending.
If scrottie had said "self-awareness and free-will help perpetuate the species", I wouldn't have asked for further clarification. I believe that statement to be true.
I likely have read too much into and too literally his comments, but it seemed to me he was saying:
1. Free will has come as a result of it being required for survival of the fittest.
I contend that free will, if equated to non-deterministic behavior, can arise in any sufficiently complex self-referencing system. Needing to survive isn't necessary for it to develop.
2. We couldn't survive without free will.
I contend that the argument, if true, has to extend to all non-extinct lifeforms and their ancestors all the way back to the first self-replicating protein.
I am not at all trying to imply that either quality doesn't help in our survival - only that neither one depend on one another exclusively.
In any event, this is not at all perl related unless we start talking about genetic algorithms. In which case, it can be shown that the fittest need not be self-aware nor have free will in order to survive.
Re:Right - except
Aristotle on 2008-05-03T02:12:44
Well, the point I was making is that free will is clearly unnecessary for survival in general, but may well have been necessary for the survival of mankind. Scott’s statements did start “why do humans have”, after all.
Re:Right - except
awwaiid on 2008-05-03T04:39:37
Yeah. What Aristotle said.Re:Can you expand on a couple of things
slanning on 2008-05-05T08:15:17
I had the same confusion as Limbic Region on these points, though I'm not "Christian".
You say it's very simple. But you make the qualification that you're not convinced we have "free will". If we do not -- that is if we have no control over our actions but are, rather, subject to deterministic laws of physics -- then it's nonsense to say that "free will" would have anything to do with "our survival", since "our survival" would be just part of a kind of inevitable unfolding of events. And if by "free will", you just mean "a sense of free will" -- that is, we feel free but in fact are not -- this argument also applies, though of course we would feel it does not.
On the other hand, if we do in fact have "free will" -- that is, we have a non-physical power separate from our bodies which can nonetheless still manage to control our physical bodies -- then you have to defend mind-body duality, an argument that I don't think has fared well over the centuries of philosophical debate on it. But I think there would in this case be some sense in saying that this property of "free will" could help prolong "our survival".
I put "our survival" in quotes, because I think this phrase is itself full of difficulties. How one defines a living being, a human being, the species of human beings, the survival of the human species - I don't think these are easy to pin down exactly. And why, for example, is the "survival" of the blobs of flesh that, taken as a whole, we identify with "the human species" more important than, say, the survival of "the mountain species" -- mountains, as a species, have clearly outlasted any of the "living beings" by far, even without any free will.
Apparently
;) Re:Can you expand on a couple of things
Aristotle on 2008-05-05T10:11:06
I agree entirely.
:-) Well, almost – I have to point out that our current understanding of physics is that there is no determinism at the quantum level. Determinism exists only at the macro scale as a probabilistic effect (essentially, there are so overwhelmingly many quantum systems interacting in any meso- or macroscopic system that all statistical deviations are wiped out in the aggregate). But since our brains are composed of parts that operate at small enough scales that quantum effects can contribute significantly, there is a possibility that the behaviour of a human truly is non-deterministic. Note further that ever since Poincaré we know that the behaviour even of completely deterministic systems like a 3-body gravitational interaction can be unpredictable.
So the subject is much more complex than meets the eye.
However, I think the problem with the idea of free will is much simpler than all of that: what does the term even mean? It doesn’t mean random action, does it? So if we don’t act randomly, how do we “freely decide” how to act? It seems obvious to me that it comes down to an accumulation of past experience, which informs our actions in the future. That is a certainly a deterministic process to some extent, if only probabilistically. It just involves so many feedback loops that the behaviour of the system is unpredictable and potentially wildly chaotic. Maybe there is a small random input due to quantum effects. So minute differences in circumstances may lead to drastically different reactions – though often only to minutely different ones.
By and large, this concept fits my experience with how humans act quite well. (Of course it’s not even close to something you could call a theory.)
