Minds and Machines

[Silent Knight]

Recently in my philosophy class we have been discussing intelligence of machines. There have been several arguments about why machines can't be intelligent and debate about how to define intelligence. Hugo de Garis has been attending our class for a while and he did a lecture on machine intelligence. He claims that if Moore's law holds there will be god-like machines by the end of this century. IIRC, Hugo predicts that by 2020 one byte of information will be the size of an electron so that something the size of a mouse could hold trillions of bytes of information.

One of the main arguments against machine intelligence is that have to be programmed. I tend to agree with this, but I don't understand computer programming. My professor says that humans are programmed too. Nature gave us the hardware and culture, education, and our parents supply the software. I said in class that you don't have to rewire a human brain to teach a person how to do something, but you have to type in information to program a computer. He said heuristic programming isn't that simple. Hugo pointed out that he encountered a talking computer when he called Dell support. I have experienced the same thing when calling a credit card company. I would like to try an experiment of using a random word such as "doorknob" to see if the computer accepts it or not.

Alan Turing came up with a test to see if you could tell the different between machine answers and computer answers. A computer is in one room and a human is in another and an interrogator asks any variety of questions to determine which room holds the machine. If the interrogator can't tell the difference then the machine passes the test. He then concluded that a machine passing this test would be considered intelligent. My professor calls this a question switcheroo fallacy, where you take the answer from one question to answer another question.

Imagine a machine such as Bicentennial Man, Data, or the kid in A.I. Would you consider them intelligent? We watched a Star Trek episode where Data wants to resign from the star fleet to avoid being experimented on and so there is a tribunal to determine whether he has rights. The question of intelligence is really fuzzy, so is the question of what is human. We accept disabled people as still being human, so to be fair should you also apply a lenient rule to machine intelligence? We can consider all people to be humans, but is there a line for intelligence and being self-aware? Terri Schiavo is human, but it's probably debatable as to whether she is self-aware and still intelligent. I see a problem with the least common denominator when addressing machines. Should I consider the elevator and my television to be intelligent? Should they be given rights?

So what is your opinion on this issue? Do you have a boundary for what you would consider an intelligent human or machine?
Hopefully this is appropriate material.

[Nicolas]

What computers/machines fail to have is an upright knowledge of their own existance/importance. They can't be egoistic.

A neural network flight computer that learns how it can still fly a damaged plane (experimental development) knows the different elements of the plane, but does not know it in itself exists. It doesn't matter for the thing if it was unplugged, or if it would be overridden by another flight computer (it does not feel concurrence, or being less important), it has no own WILL. It is as free as it is designed to be.

[Donnie B.]

It's a pretty deep question, since we aren't really sure what our consciousness is, or how it arises from the function of the brain.

However, I tend to fall on the side of the debate that says the question is already resolved, since we ourselves are examples of machines that are self-aware. The hardware is different, but there's no ghost in the organic machine either.

[Swift]

A very interesting topic that I have a couple of thoughts on.

He claims that if Moore's law holds there will be god-like machines by the end of this century. IIRC, Hugo predicts that by 2020 one byte of information will be the size of an electron so that something the size of a mouse could hold trillions of bytes of information.

Moore's Law is not a law like the Law of Relativity. Its a trend in the improvement of transistor density on chips. Its more of a marketing goal. There are serious doubts that it can continue forever. At some point the physics changes radically when the devices get very small. But on the flip side, I'm not convinced that the limit to machine intelligence is number of transistors on a chip or info per volume.

One of the main arguments against machine intelligence is that have to be programmed. I tend to agree with this, but I don't understand computer programming. My professor says that humans are programmed too.

I don't understand why this is an argument against machine intelligence (but I didn't take the course :wink: ). I agree with your professor's statement.

The question of intelligence is really fuzzy, so is the question of what is human. We accept disabled people as still being human, so to be fair should you also apply a lenient rule to machine intelligence? We can consider all people to be humans, but is there a line for intelligence and being self-aware? Terri Schiavo is human, but it's probably debatable as to whether she is self-aware and still intelligent.

I think the question of intelligence and the question of being human are somewhat related but rather different. Actually, "being human" is probably a wrong term, because in the Star Trek universe, Vulcans aren't human, nor are all those energy being. Maybe sentient is a better term.

Self-awareness is probably a necessary part of being sentient, but I don't think it is the whole story either, since my cat is almost certainly self-aware, but probably not sentient. Some interesting definitions

I don't want to get into the particulars of Terri Schiavo's case, but as far as the general question of brain-dead "people", in my opinion, I'm not convinced that they are human any longer. The only thing that remains sort of human about them is the reaction of the living to their bodies.

[Donnie B.]

... if Moore's law holds there will be god-like machines by the end of this century. IIRC, Hugo predicts that by 2020 one byte of information will be the size of an electron so that something the size of a mouse could hold trillions of bytes of information.

Incidentally, something the size of a mouse can already hold trillions of bytes of information, even if we exclude a mouse-size brain and just refer to our current technologies. No, we don't have MAM (Mouse Access Memory), but look at the size of the bits on the surface of a next-generation DVD (BluRay or similar) and extend that to three dimensions...

As to Moore's Law, it may hit a wall, in practice if not in theory, before 2020. We're already having problems shrinking to the next lower feature size (around 65nm), which is turning out to be hideously expensive to achieve. Things get worse quickly at even lower scales. The problems may be surmountable for another ten years, but we're getting close to the physical limits of silicon. Make the features small enough and quantum effects start to rear their ugly heads.

[Nicolas]

Your cat may not be sentient, but she certainly is intelligent (and inherently insane as stated earlier ).

Cats certainly are self-aware. They strongly feel good-being, allowance, attention and concurrence.

[Silent Knight]

One of the main arguments against machine intelligence is that have to be programmed. I tend to agree with this, but I don't understand computer programming. My professor says that humans are programmed too.

I don't understand why this is an argument against machine intelligence (but I didn't take the course :wink: ). I agree with your professor's statement.

Let me try to clarify.. I think the argument is basically that machines can only do what they are programmed to do. They can't deviate from the programming, whereas humans can do a wide variety of things without being told they can or can't. But my professor is saying that is a nonsense argument because humans are also programmed.

I think the argument comes from Lady Lovelace who basically said that a computer can only do what it is programmed to do.

[A Thousand Pardons]

Should I consider the elevator and my television to be intelligent? Should they be given rights?

Careful, you're about to flunk your Turing test.

[electromagneticpulse]

Okay i was going to make a big long post on this but i don't want to waist peoples time reading it

I would have to say AI will be moronic. Okay say the 2020 thing would be when computers would have our "computing" power. They would still be dumb.

I have several process centres in my brain. Left side and right side all my senses etc, all of which are processed in seperate parts of the brain. The bigger problem is that its lots of little processors in the brain each interlinked that can work things out. So technology would have to come a long long way to get that but then we have hormones etc which also work it out.

Without hormones no emotions, no terminators or matrix robots. They would have no fear of death and no reason to retaliate when threatened. I don't see the aim of discussing it in a philosophy class, try debating if there is a point in having philosophy class