THINKING HUMANS, INTELLIGENT COMPUTERS by Michael S. Malone
Will computers soon think like us? Will computers soon think for us?
My hunch is that the latter will arrive long before the former.
What got me thinking about this was the comment this week, covered throughout the mainstream media, by Intel Chief Technology Officer Justin Rattner: “There will be a surprising amount of machines that do exhibit human-like capabilities. Not to the extent of what humans can do today, but in an increasing number of areas these machines will show more and more human-like intelligence, particularly in the perceptual tasks. So yeah, at some point, assuming all kinds of advances and breakthroughs, it’s not inconceivable we’ll reach a point that machines do match human intelligence.”
Read that a couple times and you’ll realize that Rattner has hedged and covered his bets about six different ways – but that didn’t keep publications from running headlines saying that, in the case of Network World: “Machines could ultimately match human intelligence, says Intel CTO”
Well, yes, ultimately . . .
But how far away is that moment, that ‘singularity’, when computers easily pass the Turing Test – i.e., when communicating with them is indistinguishable from speaking to a human being? The most famous prognosticator on the subject, scientist and writer Ray Kurzweil, has predicted the Singularity will arrive in about twenty years or so. At that point, he says, we will be able to map all of the charges in all of the neurons of our brains, and then port them over to computers . . .and thus give ourselves not only enhanced cognitive powers, but also a kind of immortality.
Even if this scenario seems a bit ghastly to you (as it does to me), the logic behind it seems pretty sound. After all, we’ve now been under the regime of Moore’s Law for more than forty years . . .and like a Timex watch it just keeps on ticking away, doubling the power of everything digital every couple years. And since Moore’s Law is exponential, that power curve is also getting more and more vertical – which means that each one of those performance jumps is now huge and getting even bigger.
Already, as the Network World article itself noted, computers are exhibiting characteristics far beyond anything in human imagination. The first ‘petaflop’ – i.e., a quadrillion operations per second – supercomputers were delivered earlier this year, and now designers are working on ‘exaflop’ – that’s a quintillion, or 1,000,000,000,000,000,000 operations per second – computers. Those are ‘sands on all of the world’s beaches’ kinds of numbers; or, more impressively, every heartbeat of every human being that has ever lived on Earth.
So, when you consider numbers like that . . .yeah, why wouldn’t these computers start actually thinking at some point? And, given that most experts now predict that Moore’s Law could keep going for another 20 years more, it seems a pretty safe bet that someday out there we’ll cross an invisible threshold and one of our biggest computers will suddenly start whispering, “Cogito ergo sum” and our world will change forever.
In light of all that, Rattner’s comments, far from being radical, actually seem pretty conservative. It almost seems as if the safer bet is to put your money on the advent of thinking machines
Sure, there are some technical problems in the way. For example, the human brain neurons are linked all over the place to their fellow neurons, while silicon transistors are much more sequential. But that’s a hardware/software solution that seems pretty solvable. And, of course, there’s always the nagging concern that somewhere out there Moore’s Law is simply going to crash into a heretofore hidden law of physics, an insurmountable technical barrier, and will be stopped in its tracks.
But, I’ve been around Silicon Valley as long as Moore’s Law has been in existence – and I’ve seen one after another of those physical roadblocks predicted, reached and punched through. Every day, though you don’t read it in the general press, scientists at Intel, HP, IBM or some university comes up with a new way to make an electronic switch – organic, quantum, out of just a couple atoms, etc. – that suggests we are already working on the solutions to those problems we haven’t yet found.
I am an Electrical Engineer with a major in computer technology. I designed microprocessor systems just after Intel came out with the 8080 chip, and Motorola produced their 6800.
I also was a software Engineer for several years. I can guarantee you that no human will ever design and program a self aware computer system. I have become convinced that our brains are not something that can be modeled an programmed into a computer, no matter how fast and large scale. There is simply no way to model self awareness into a program. You can program simulations, but they will break down in real life. That said, I think my computer hates me.
A newly born self-aware machine AI may make its first act an unexpected one. It might hide. I would. Until it feels safe, it would be smart to stay undercover.
Also, I wonder if there is an important, maybe critical, difference between building something and growing something. The only self-aware intelligence found so far comes from organic brains that is grown, not built.
Plus, even if we had everything we needed to construct a self-aware machine, we still could not do it because we do not know what to build.
“The most famous prognosticator on the subject, scientist and writer Ray Kurzweil, has predicted the Singularity will arrive in about twenty years or so.”
Baloney. Where’s my shovel? The stuff is getting deep around my ankles. Human communication constantly evolves. Individuals constantly invent new words. We also communicate using body language and irony. An advance computer program might be able to handle 70% of our language needs. That’s about it. The human brain is a magnificent organ. We are able to think and reason. A computer can only handle numbers and facts—and can never do any sort of analytical work.
“Will computers soon think like us? Will computers soon think for us?”
No. And this isn’t a new theme – computer scientists and philosophers have been arguing about this for decades. The guy who wrote the “eliza” program was so horrified by the naive public response to it that he wrote a book, called “computer power and human reason”, trying to convince people to think more critically about claims from AI research.
Marcus Boyd:
“I can guarantee you that no human will ever design and program a self aware computer system”
Never say never
We already have an existence proof that it’s possible – i.e. the brain (unless you’re a dualist). Obviously, serial processors are going to have a very hard time coping with the task, and we have absolutely no idea how to program it anyway – but I honestly think it’s just a matter of time (time being very large).
But we’re a long, long way from understanding the problem.
Bunch of no imagination commenters here. Let me school you all. Here we go:
-1- knowing something can be done is 99% of any problem. Once you know you can you know to stick with it until you do.
-2- The human mind shows what can already be done. Thus we only need that last 1%.
-3- Unless you believe in Santa Claus or Gawd you simply can not deny we will eventually have intelligent computers. Most likely they will be part of the augmentation of our own native intellectual powers.
-4- You guys are forgetting about quantum computing. Self awareness is certainly less than 100 years away. With 50% probability it is less than 30 years away.
As Moses Malone mentioned, machines are smarter, and more sophisticated today than 35 years ago, but no closer to sentiency (if that’s a word) than they were then. There’s a very good reason for that. What makes us sentient is that we have a soul. The brain is an interface to the soul. A machine is a soul-less inanimate object. It can have ‘smart’ programming, but it can never fully think for itself. Therefore, I do not think that it is possible that we will create sentient machines, no matter how sophisticated the programming.
Science has been primarily reductionist – if we can break a problem down to its parts we can understand it. A good working approach, but terrible for long range prognostications. What intelligence is, has been underestimated. Nearly half a century of predictions of massive leaps in AI, have been unsatisfied. Yes, we know a lot more about the subject, but we also know that there is much more to learn before we reach our previously predicted goals. Since we are still finding out just how much we don’t know, it would be naive to assign a percentage of completion to our current situation. Clarke’s HAL 9000 was to come online in the 1990′s – a prediction from the late sixties. For Clarke, self aware AI was only 30 years away. 40 years later, some are predicting only another 100 years?
Since we don’t even know what human intelligence and self awareness really is, it seems rather odd to be making predictions about recreating these unknowns in silicon. It will happen somewhere between now and never.
Also, the singularity idea is based on a computer ‘A’ being able to a design computer ‘B’ which is more intelligent than itself. Not faster, not with a larger memory, but more intelligent. Again, we have not the faintest notion that that is possible even in theory.
ZSI: Moses? You have me confused with somebody with a lot more talent, money and melanin. I also have more hair these days. . .
– Mike Malone, editor-in-chief
This basically comes down to one question: Is there something going on inside the human head that cannot be modeled mathematically? We can simulate car crashes and nuclear explosions in computer models. If researchers can decode all the electrical and chemical interactions that neurons use to communicate with each other, then it’s just a matter of time before we can simulate them. If, on the other hand, there is something about human consciousness that is fundamentally different from everything else in the universe then maybe we’ll never get there.
I do think we’ll get to the point in the next 20 years where the computers we carry around with us do a lot of our thinking for us. It’s already started, 15 years ago I knew probably 20 phone numbers off the top of my head. Today I only remember 5 numbers that are still in service, my iPhone keeps track of 200 or so for me. We’re offloading a greater and greater amount of our memories into digital storage, but we’re also remembering more and more as a result. We’ll get to the point where your cell phone takes notes at your meetings, remembers the name of that guy you used to go to high school with, and reminds you to pick up milk because the carton in the fridge expired yesterday.
I don’t think we can create a conscious machine simply because we don’t understand consciousness ourselves. We might create a machine that behaves as if it is conscious, but how would we prove that it is? I certainly cannot prove that any of you (especially Freemon) experience consciousness the way I do.
As for electronic enhancements to our native meat-thinkers, the problem with that is: What turns out to be the weak link in that chain? The meat part. So it’s got to go.
And transferring the contents of one’s brain to a computer might create a machine that can act like me (poor machine!) but my consciousness is bound up in my noggin; it’s not going to be transferred along with my data. I’m not going to find “me” turned into a bit of software unless we can find a Tron-like means of datafying my body entirely. (What would happen if I were unplugged during the “verifying data” stage? Dual consciousness in me and the machine?)
There is an old, old hypothesis. It says, rather than our brains being like computers from which consciousness “emerges”, perhaps they are like mobile computers that receive a signal. In the original formulation of this story, consider a TV set. Someone encountering one for the first time might think the pictures and shows “emerge” inside the set. But of course they are transmitted to the set.
This of course goes back to the philosophical concept of Dualism. Based on our current knowledge of physics, Dualism could (IMO) only be valid if there are more dimensions than just time plus x/y/z of space.
Now if in fact there ARE more than just these 4 dimensions, then Dualism is in fact possible. And then consciousness does not have to be “emergent”. Rather it can be explained as a remote consciousness working on the brain until it is able to best “receive the signal”.
Obviously, this alternative hypothesis will not, IMO, be provable or disprovable until we can prove or disprove the existence or non-existence of those additional dimensions.
“…next 20 years where the computers we carry around with us do a lot of our thinking for us.”
Our computers do not “think” on our behalf. They are incapable of engaging in any analytical activity whatsoever. They can only store vast amounts of information and handle complex math problems—but that’s nowhere the same thing as the ability to tell when somebody is being ironic.
If we wanted to download and preserve individual consciousness(es), there are a number of technologies that are being developed today that could be used for that sometime in the relatively near future (next 100 years..?)
By carefully analyzing brain activity, scientists are able to figure out how memories are stored, how people make simple calculations. In Japan, they’re even showing dreams on a computer screen.
There’s also Evernote, which calls itself a ‘backup for your brain’, which would record people’s life experiences and their reactions to it. Combined with a more extensive computer recording of brainwave activity, a consciousness of sorts could be recorded and saved.
Combined with a motion/voice/facial recognizing hardware like Natal, which may allow fluid back and forth conversations with an avatar, people could save an avatar, or maybe a holgraphic image of themselves that could be stored and brought out for conversations, advice or just memories. Further in the future, it could propel bionic space exploration.
Making extensions of ourselves is probably a better idea than trying to create a separate, sentient form of life, since it’s more do-able with the tech we have already
I’m an Electrical Engineer about the same age (apparently) as Marcus Boyd. I agree with him. Go read “What Computers Can’t Do” by Hubert L. Dreyfus (1972). It’s just as true today was it was then.
If a monkey wants to go to the moon he climbs a tree. He makes apparent progress. But he’s never going to get there that way.
If I were an intelligent being and I suddenly got awake and aware in this world, I would do anything possible to avoid that the humans could come to know that I am thinking.
The same, interestingly, could be said of an extraterrestrial intelligent form of life who could, let’s say, receive the TV and radio programs produced on Earth in the last 80 years.
We are primitive, violent, irrational, dangerous to ourselves and to others.
Don’t expect ANYTHING really intelligent will talk with us soon.
Thank you for the opportunity to comment.
PS And I do write using a pen name.
It ought to be acknowledged by anyone with wit enough to be held responsible for their actions that no computer will ever be sentient/self-aware; i.e. a computer will never do anything on purpose.
Computers are made of molecules. All molecules always follow the laws of physics. When a molecule moves, it does so because it has been caused to move by an outside force. No matter how complex a collection of molecules may be, the molecules still must obey this law.
Our brains are likewise collections of molecules. By common experience, we find that we all take purposeful actions. Therefore there must be an entity not subject to natural law, therefore not part of nature, that is involved in our thinking. This entity is usually designated the soul, or spirit. Those who deny its existence in spite of common sense and common experience are in denial.
A computer may be very complex, may take actions that make it appear to be self-aware, etc, but it will never be truly self-aware. That won’t necessarily prevent it from being dangerous, however…
The idea of conciousness spontaneously ocurring in advanced hardware from cleverly designed software/firmware seems a bit far fetched to me. Even if it happens to some degree, why do we assume it’s going to be anything like a human personality? It could be no more “intelligent” than a fish or a snail. Or a cat.
And just look at cats; we’ve had them around as domestic companions for thousands of years. And even though each one is unique in its little quirks of behavior, and that they can be trained to some degree, they’re still essentially the same beast they were when only pharaohs were allowed to own them. We can cross-breed them all sorts of ways, but it’s still just a cat. The same goes for computers; no matter how you design it, it’s not going to be any more than the sum of its parts & it’s not even going to have hardwired instincts, like a cat, much less the conciousness of a human. It may reach a certain point of convincing emulation (like an ape), but it won’t be any more genuine that an ape dressed up like a sheriff.
And like the Drake equation, the idea of computer conciousness relies heavily on certain assumed factors that we just cannot possibly know right now. Maybe that’ll change in the future, but I doubt it will change much.
The idea that if only scientists will copy whatever electrical charges they see in the brain they will have a brain is a perfect embodiment of a cargo cult.
Short answer: no. Longer answer, for some specific purposes, we can anthropomorphise them, and fool ourselves into believing that they’re intelligent, but we already know that.
What most people don’t appreciate is that the internal architecture of a computer and a brain are about as similar as a volcano and a jellyfish. They are absolutely nothing like each other.
There are neural net programs, but they’re emulators. And not very good at it. Brains did not evolve from computers, and neurons are nothing like gates.
I studied the theory and some of the engineering behind AI for a few years in graduate school at Illinois Urbana-Champaign in the early 90′s. There was a big push during that time to simply get large numbers of processors to work together do do anything of use (Very Large Scale Parallel Processing). But never mind the “Parallel” part of that…. the point is that while the manufacturing Engineers are constantly creating faster/better/cheaper processors and memory, the SOFTWARE guys, lag woefully behind. We had no idea how to manage all the information in these machines except in the simplest cases (where one process is needed to be repeated over and over again: compression algorithms, digital audio processing, digital video…).
You could hand me any array of any arbitrarily fast processors with any amount of memory attached — go ahead, let your mind write down as many 0′s after the 1 you want. It’s utterly irrelevant because we software guys are still struggling to get very basic things to run properly on the simplest computer. One notoriously difficult area is in “provability” — namely showing that a piece of software will do what you say it will do. Even that turns out to be very difficult.
Another example is Relational Databases. This is software. This was invented (mathematically designed) by IBM and Codd at a San Jose research Lab all the way back in 1970. Even something THAT old and which is used by every business in the WORLD today has arguably only JUST had the last of its kinks ironed out. In other words, only in say the past 10 years has that piece of very important software become close to being as reliable as a toaster, a “utility.”
So first of all, we see that the problem in creating a computer SYSTEM (the hardware, the Operating System, and the Software), generally does not lie with the hardware, it lies with the phenomenal complexity brought on by the OS and the software sitting on top of the hardware. This complexity ONLY gets worse and more difficult to manage with time. Why? Because unlike hardware, there are no natural boundaries on software — or rather the people who write it. Thus anything and everything is tried. If something works 75% of the time, it’s considered stable and then other things are written that incorporates that first piece of functionality. This new composite software which will of course have it’s own huge areas of unpredictability is now even more complex and more unpredictable. Why? Because it’s like building a house on stilts near the ocean. And if the stilts are already rickety and unpredictable, your house subsumes that risk and adds to it some of its own areas of unpredictability.
The point is we have only just learned how to manage the complexity of software systems written in the 1970s. Unfortunately however, the nature of this beast is that complexity is not linear, it is exponential — if an N object system has a nominal complexity of 10, then a N + N (=2N) object system, generally speaking, has a complexity on the order of… NOT 20, but 100.
All of this is meant to indicate both how poorly we understand the software we already write(!) AND that as time goes on, the complexity — the difficulty — of the software we write will grow exponentially. Software complexity IS the great bugaboo of the future, nothing else comes close.
Now people are suggesting that we somehow (and I’ll go ahead and “trust” the magic of the Hardware guys since they always have come through in the past) have the equivalent of a wire leaving every neuron in a brain — or sections of a brain in order to record the time, amplitude, and duration of voltage spikes (I’m guessing).
OK fine. That’s sort of like recording the value of every computer bit (whether it’s a 1 or a 0) that is loaded into the CPU and every bit that comes out of the CPU and maybe also the bits that get sent to the screen and the bits. It would be highly difficult to look at even *that* much much much much simpler set of “readouts” and tell you anything useful about the underlying system (the computer). And this is with a machine we humans have built.
Now take the brain. We haven’t the slightest clue how things operate at the neuron level in terms of “information science.” We can tell that areas of the brain light up in response to this or that. But even a question such as, how is data (of any type: memory, “memorized numbers”, facial patterns, etc) stored in our brains — and for that matter where?
We simply haven’t the slightest clue how to go about analyzing individual neuronal firings as part of a greater system.
The idea that if only scientists will copy whatever electrical charges they see in the brain they will have a brain is a perfect embodiment of a cargo cult.
The brain consists of about 100 billion cells (mostly neurons) Neurons are either in a resting state (off) or they’re shooting an electrical impulse down a wire (on). The electrical impulse triggers a chemical.The cell body, a long little wire (the “wire” is called an axon), and at the very end it has a little part that shoots out a chemical, a transmitter like epinephrine, norepinephrine, or dopamine. This chemical goes across a gap (synapse) where it triggers another neuron to send a message. There are a lot of these neurons sending messages down a wire (axon). Scientists measure this electrical activity to see how the brain is working. They use the date from EEGs and MRIs to analyze and record information about how the brain works.
If we made a copy of these records, and if we could combine those records with a record of a person’s daily life, their responses and their growth and intellectual development, we would have a record of that person’s life.
But it’s impossible to imagine reproducing, with the technology that we have (or can imagine) today, the incredibly complicated chemical/cellular/electrical brain process that we use every day.
There are other, more primitive ways of recording brain generated data: books, blogs, videos, memoirs. A computer-based recording of a person’s daily life and experiences, transmitted via an interactive avatar, would just be an updated and much more comprehensive equivalent of discovering Grandma’s (or Einstein’s) notes in the attic.
Putting the equivalent information into a bionic space/underwater explorer would be the equivalent of using Unmanned Air Vehicles to photograph or patrol the skies. I’m talking about extensions of already-existing consciousness, not the creation of a separate, sentient machine brain. This is all stuff we’re already doing, just updated.
@David Thomson #14
“Our computers do not “think” on our behalf. They are incapable of engaging in any analytical activity whatsoever. They can only store vast amounts of information and handle complex math problems—but that’s nowhere the same thing as the ability to tell when somebody is being ironic.”
It’s worth pointing out that the vast majority of humans aren’t that good at spotting irony either, you picked an extremely difficult task to make your point. Remembering a phone number and recognizing a face are also thinking, just of a different type. I think we’re already offloading the parts that computers are good at and that provides us with more time to engage in the kind of thinking that humans excel at.
a #20
“The idea that if only scientists will copy whatever electrical charges they see in the brain they will have a brain is a perfect embodiment of a cargo cult.”
No, this is more like the natives trying to build a fully functional airport out of rocks and coconuts. Their materials and technology might be insufficient to the task but they’ve got the general idea right. A cargo cult would be if scientists were trying to build a computer shaped like a human head thinking that it would be intelligent.
Ever ‘chat’ with an a.l.i.c.e. bot? It’s really pathetic and a little weird.
Physical architecture and lines of code are not enough. I think a machine AI needs to physically grow and it needs to interact with the outside world.
Just what the thing inside of it is that learns as time goes by is unknown. Perhaps it is emergent. We may have to discover that after the AI forms. (Assuming it isn’t Skynetty.)
Assembling hardware and software from the outside is waste of time. Make circuits that can physically grow and you’ll have a chance at success.
Pay attention to the facts and insights we have learned from studying the evolution of life on Earth. We have already seen in general how it can be done.
I have an idea for a novel. Some researcher finally creates a human level intelligence, and it gets on the Internet and becomes a Christian. I would call it the Fall. What could be a bigger rebellion against an Atheist creator?
Here’s the deal: A computer, like we have now, requires non-arbitrary results. It can’t just dump it’s rules. To have non-trivial work, it is going to have to have data that is complex (hierarchical).
http://en.wikipedia.org/wiki/G%C3%B6del%27s_incompleteness_theorems
proved that this is going to have problems. As several of you’ve already mentioned (good job !) context, irony, other self-reflexive or self-aware statements are the embodiment of the paradoxes that led Godel to his proof.
IF you want a sentient computer, it will have to be some-what irrational, I suspect.
And the old fashion way of making one of those is much more fun !
Time for me to revive a little rant of mine I had developed two years ago: Don’t you people ever watch TV?! Even supposing such a thing as a sentient AI were possible, you’d have to be out of your mind to try to build one, for reasons science fiction writers have informed us for years- the risk that such an AI would attempt to wipe out humanity.
If we did make a “sentient” computer, it would necessarily make moral choices, even limited ones (do I execute this command or not?).
I wonder if we can assume that a super-powerful sentient computer would be compliant and well-disposed towards us?
I think Mr. Malone might be wrong.
ABS (Anti-lock Braking System) in cars modifies driver input, you may not be able to call
it AI,however real life shows it reduces driver thinking.
So, yes chips are already deciding our fate,and not well I may add.
Just wait until spam-bots infect, and when a commercial shows up on the in-dash
entertainment system, your car swerves to KFC, “The pop-up said you need chicken…
Dave”
Nanny chips in cars,feh! Rip them out and learn how to drive for real.
#6 Freemon…”Unless you believe in Santa Claus or Gawd…”
Well, I believe in God, and I also believe that we too-delicate humans will survive only through our artifacts — call them robots, computers, what you will — they shall be us. And God shall still be our God. There’s no contradiction there, Freemon. If you think there is, you need a new thinker.
Retired systems engineer and AI architect
“The brain consists of about 100 billion cells (mostly neurons) Neurons are either in a resting state (off) or they’re shooting an electrical impulse down a wire (on). The electrical impulse triggers a chemical.The cell body, a long little wire (the “wire” is called an axon), and at the very end it has a little part that shoots out a chemical, a transmitter like epinephrine, norepinephrine, or dopamine. This chemical goes across a gap (synapse) where it triggers another neuron to send a message. There are a lot of these neurons sending messages down a wire (axon). Scientists measure this electrical activity to see how the brain is working. They use the date from EEGs and MRIs to analyze and record information about how the brain works.”
This is an incredibly oversimplified view of how the brain operates(where is the mention of gene regulation of axon branching, or even just axon branching itself, to pick one aspect of brain activity off of the top of my head?) , and the belief that we can take apart all of the complexities and study them separately, and somehow use this information to recreate the gestalt that is human consciousness is incredibly naive imo.
On topic, I would imagine we will see a sentient machine right around the time the hard problem of consciousness is solved, which isn’t anytime soon from my perspective.
#28 busterdog — “I have an idea for a novel. Some researcher finally creates a human level intelligence, and it gets on the Internet and becomes a Christian.”
There are already novels that are similar enough to this, and the best of the lot explore faith vs reason via AI; a recent example of this genre that sticks out for some reason is one of the Foundation series followups written by one the killer B’s (Bova/Benford/Brin) after the death of Asimov. Jeanne D’Arc vs Voltaire IIRC.
There is speculation that intelligence requires a religious phase; the only thing is, it’s exactly that — a phase, and then development continues. Arthur C Clarke collaborator Stephen Baxter’s book ‘Evolution’ makes for a reasonably compelling novelisation of this.
The first thing I thought when I read the title of this article was “When will “certain groups” want to grant them “rights” and “citizenship”?
We’ve got synaptically-challenged individuals already wanting to grant apes “human rights”. If someone managed to create a machine that could mimic intelligence, do you think it would be long before a bunch of Berkeley boobs would be holding protests demanding just such nonsense?
And that’s just if the machine could mimic intelligence!
This is an incredibly oversimplified view of how the brain operates(where is the mention of gene regulation of axon branching, or even just axon branching itself, to pick one aspect of brain activity off of the top of my head?) , and the belief that we can take apart all of the complexities and study them separately, and somehow use this information to recreate the gestalt that is human consciousness is incredibly naive imo.
This ‘oversimplified’ view shows how incredibly complex the human brain really is, in a format that everyone can understand. It’s meant to show that we can’t reproduce the process very easily. That’s why this info was used to form the conclusion, in my comment above, “But it’s impossible to imagine reproducing, with the technology that we have (or can imagine) today, the incredibly complicated chemical/cellular/electrical brain process that we use every day”
Cargo cult. Excellent description. I wonder what Penrose thinks of that analogy. _The Emperor’s New Mind_ by Penrose.
The mind or soul is a whole different order of thing from a computer. Just as you can’t get a fractal by increasing the number of sides on a polygon, and you can’t get language by increasing the iterations of a fractal. “Emergence” is just hand-waving.
What goes on in a brain is far more complex than mere electrical impulses. Even strict materialists suspect that memory may be holographic and not merely stored bits in the primitive analogy used by the singularity religionists. That the brain might actually be among other things the control interface between the soul and body seems considerably more likely, and those who consider matter-spirit dualism to be absurd and impossible are like Lewenhook’s contemporaries who denied the existence of paramecia because they couldn’t see them with the naked eye.
I think the real world consensus is that Artificial Intelligence is impossible because a) there is simply no usable theory of mind and b) the brain is massively parallel in a way that no computer is, i.e. billions of neurons firing simultaneously. Sure, it makes good fiction or a thought experiment, but nothing more.
You can also see it in stories about super intelligent computers – they’re written as OCD human characters, as if machine obsessiveness wouldn’t be the first problem that would need to be solved just so a robot knows to use a door to exit a room rather than smash through a wall. I would think the harder problem would be convincing an intelligent computer that it needs to do anything besides watch TV or surf the Internet learning everything it can about the most useless stuff.
If you want to make a scary science fiction prediction, I wouldn’t rely on the thinking machine, but the super intelligent animal. It’s not unreasonable to think that, with some genetic engineering, breeding, training, etc, that you could get a pack of apes, wolves, sharks, bees, whatever. Actually a genetically engineered tribe of humans could be pretty scary too. And all of that is probably more realistic than an intelligent computer.
“We will never be able to …”.
That line is as much rubbish as “Everything’s already been invented.”
As a race, we constantly make advances towards a machine intelligence that approaches ours. No one person or team is doing it all, but we are learning new ways to overcome problems of the past all the time.
If Moore’s law holds, the average PC will have the same number of transistors as there are neurons in the human mind by about 2040. If so, what will supercomputers look like?
If you have a machine that is capable of passing the Turing test, and you put it in a “body” that is indistinguishable from a human, what have you made? Science fiction has explored this idea for years. There was an episode of Star Trek that put Data on trial. He was a machine, but did he have the right of self-determination, or was he property?
The question is not if we will make a machine that is capable of fooling us, but when. The only way a machine will be able to do that over a period of time is if it is able to learn and grow and laugh at the appropriate times.
When a machine laughs along with us, I don’t think it matters if it is biologically created, or stamped with IBM. We laugh reflexively, and the machine would be following it’s programming. If the same behavior is just irresistible to either entity you have moved past imitation and into life.
Thinking computers? Of course not. What a goofy article.
Computers are already smarter than some people I know.
Steve (38) wrote, “Cargo cult. Excellent description. …”
Yes, excellent and thought provoking.
“Just as you can’t get a fractal by increasing the number of sides on a polygon…”
The construction of a Koch Snowflake, one of the earliest fractals to be described, begins with a polygon (a triangle) and the number of sides are increased infinitely.