[onoway]

Luddites are out of luck; the pandora's box of technology has been flung wide open and cannot again be tucked away. A couple of TED talks which I found very sobering
http://www.ted.com/t...ngineering.html and
http://www.ted.com/t...ots_of_war.html

[kenberg]

Two very interesting talks.

[barmar]

Scarabin, on 2013-February-05, 23:42, said:

The "technological singularity" may already be upon us: Barmar transferred the robot posts to this topic but the software left behind my post #180 which first highjacked the "irony" thread and transferred P_Marlowe #31 which threatens to morph this thread. I do not know if this is an example of robot whimsy or value judgment but perhaps we should start a thread on:"Are robots in favour of censorship?" while we still can?

I had to decide which posts to transfer to the new thread. If mistakes were made, they were due to flaws in my intelligence, not artificial intelligence.

I think I chose to leave your post #180 alone because it was about drones being used to attack people, which I thought fit the original topic more than the general discussion of robots.

[Scarabin]

barmar, on 2013-February-06, 13:00, said:

I had to decide which posts to transfer to the new thread. If mistakes were made, they were due to flaws in my intelligence, not artificial intelligence.

I think I chose to leave your post #180 alone because it was about drones being used to attack people, which I thought fit the original topic more than the general discussion of robots.

Sorry, could not resist the opportunity. Glad we can blame Ken for hijacking the "irony" thread and not me!

[Scarabin]

mike777, on 2013-February-06, 00:13, said:

It sounds like either of your tests would be a fair one as long as we can measure and compare it to a human.

For starters you need to present a standard method to measure.

My guess is a computer/bot would measure zero today.

I don't think I can suggest an adequate way of measuring my proposed tests. The trouble is that once a standard is set someone will program a computer to meet it even though the program may be "unintelligent". I think Turing changed his original test to make it measurable.

I fear I can only say I can recognize intelligence but not define it.

[barmar]

Scarabin, on 2013-February-07, 15:46, said:

I don't think I can suggest an adequate way of measuring my proposed tests. The trouble is that once a standard is set someone will program a computer to meet it even though the program may be "unintelligent". I think Turing changed his original test to make it measurable.

That's a common problem. Humans have an intuitive notion that they're "special". But every now and then we discover animals with one of abilities we considered uniquely human, and we have to move the bar. The same thing happens when we program computers to mimic some of our capabilities.

The best I can think of is that there's no single ability that makes something "intelligent". What sets us apart is the collection of a number of abilities, generally related to mental and social skills (physically, we're pretty mediocre, although bipedality was a major contributor to our evolution of intelligent actions, such as tool use).

Also, there's a quantitative difference. We find animals using tools, animals thinking symbolically, animals using language more sophisticated than just grunts. But in all cases, these abilities are very limited compared to humans. We use tools all the time, animals mostly resort to them only when really necessary.

[onoway]

Is intelligence strictly a human attribute or can for instance an animal possess intelligence as well? If so, how is that demonstrated?

If intelligence is considered to be something only humans can have then there is little point in debating whether or not computers might ever be intelligent.

[barmar]

onoway, on 2013-February-07, 20:02, said:

Is intelligence strictly a human attribute or can for instance an animal possess intelligence as well? If so, how is that demonstrated?

If intelligence is considered to be something only humans can have then there is little point in debating whether or not computers might ever be intelligent.

That's the $64,000 question. What often happens is that when we teach animals or computers to do something, we redefine intelligence not to include that. So intelligence is effectively "what humans can do that animals and computers can't", and artificial intelligence is defined as an oxymoron.

I don't think there's anything that makes artificial intelligence fundamentally impossible -- the brain and body are just physical mechanisms. However, doing it may be incredibly difficult, and possibly infeasible.

It could be like fusion reactors. We know that fusion power is possible, since it's what powers stars. But power plants have many constraints that stars don't -- they have to be near humans without killing us, they have to be controllable, they need to work with small amounts of fuel, etc.

Similarly, creating intelligent devices means that they must be designed within the constraints of our ability to build things. When we build things, we need them to operate by rules we understand. Evolution was under no such constraints, it just let the pieces fall where they might, and adapted things that worked.

Things are improving. We've been learning more and more about how the mind works, and this gives computer scientists ideas about how they can build robots. But we're also learning that it's incredibly complex, so it's not so easy to apply what we learn to artificial devices. Ignorance about how human (and even animal and plant) intelligence operates was the reason AI researchers were so hopeful that they'd be able to recreate it 50 years ago. The more we learn, the more we learn how hard it is to replicate.

[kenberg]

Scarabin, on 2013-February-06, 16:25, said:

Sorry, could not resist the opportunity. Glad we can blame Ken for hijacking the "irony" thread and not me!

What was it Nixon said? I accept the responsibility for Watergate but not the blame? Or did he accept the blame but not the responsibility?

Maybe I need to look elsewhere for a role model.

[dwar0123]

barmar, on 2013-February-07, 20:17, said:

"what humans can do that animals and computers can't", and artificial intelligence is defined as an oxymoron.

How does that logic hold up if computers can do everything(intellectually) that humans can do?

[onoway]

dwar0123, on 2013-February-08, 11:57, said:

How does that logic hold up if computers can do everything(intellectually) that humans can do?

If intelligence is by definition something only humans can possess then what computers can do requires a different definition. It seems to me to be about as valid as the arguments denying evolution but people do hang on to the idea that humans are uniquely above/apart from the rabble of the natural world and so require special rules.

It's how we justify doing what we do to the environment, for one thing.

Not that I will see it but it will be interesting to see what happens when people do develop self repairing and self replicating intelligent robots, especially if that superiority mindset is unconsciously or otherwise part of the original programming. Logic already indicates the earth would be a lot better off without humans. So then?

Some time ago I read an article which suggested we are developing our own replacement and will as a result suffer the same fate as dinosaurs, the difference being we are creating and celebrating the comet which will lead to our own extinction. To declare it cannot happen to me smells of the same arrogance that declared the Titanic unsinkable..but I have nothing to offer as a way to avoid it eventually. Eat, drink and be merry ...and beware of computers singing "daisy daisy..."

[dwar0123]

The question is a semantic illusion.

If intelligence is defined as something only humans can possess. Than intelligence is nothing.

[Scarabin]

I think I agree with Barmar's analysis, #46 & #48, although I probably disagree on minor points. The devil is in the details.

For instance, I am not sure if Barmar implies That we "move the bar" through vanity or if posts #52 & #52 introduced this idea. In either case I think we do this because we find our previous criterion no longer valid.

My previous two points really boil down to requiring a computer to be able to think like a human being, and this has to be perceptual, qualitative rather than quantitative. Let me try to clarify my thinking with a trivial example:

Suppose I adopt the test that a computer must be capable of learning, and someone writes a simple simulation of tossing a coin and the computer reports the probability of "heads" as 47% based on the frequency of ocurrence in a sample. Now the author comes to me and says his computer is deriving and learning probabilities just like a human.

I counter that the computer has not done anything not included in the original program, and that I would have been more impressed if it had commented on the futility of the experiment or had shown some "human" understanding of the results. Lo and behold, he goes away and builds in a range of statements to do just that.

Take another example, Bridgemaster: now Bridgemaster replicates human thought on a single particular problem. If it could do this for any and every bridge problem, I would accept computers had come close to intelligent.

[onoway]

Scarabin, on 2013-February-08, 20:35, said:

Take another example, Bridgemaster: now Bridgemaster replicates human thought on a single particular problem. If it could do this for any and every bridge problem, I would accept computers had come close to intelligent.

Having absolutely frustrated/humiliated myself innumerable times with Bridgemaster I don't think I like where that thought leads!

[mike777]

1) lets for starters agree the definition of intelligence and how to test and measure it as generally accepted.
2) apply it to computers over the next 20 years or whatever.
3) if the general measument changes....great.
4) if you want to assert that computers only mimic or copy intelligence ok.
5) fwiw I dont think computers pass a general test of measurement, today.
6) I sidestep the entire question of can a computer be self aware, alive or be conscious.
7) vast advances in the research and understanding of how the brain works will be needed to take the next step to an intelligent computer.

As I stated I have talked with many computer sci profs who all ..all agree that there is nothing in the laws of science that says a computer cannot be intelligent. However there is wide disagreement on how long that will take.

If it takes only 17 years lets discuss it ...if 17,000 years or longer perhaps not.

[Scarabin]

onoway, on 2013-February-08, 21:52, said:

Having absolutely frustrated/humiliated myself innumerable times with Bridgemaster I don't think I like where that thought leads!

You're probably doing better than me: I breezed through A levels 1 to 3 and only encountered difficulty at level 4 (at least that's how I remember it) but my intelligence seems to be shrinking since I have difficulties with level 3 on the later refill deals. I'm clinging to the memory that Inquiry said something about an A4 deal which I mentioned in a "State of the art" topic would only have been level 3 in the later refills!

[Scarabin]

mike777, on 2013-February-08, 22:21, said:

1) lets for starters agree the definition of intelligence and how to test and measure it as generally accepted.
2) apply it to computers over the next 20 years or whatever.
3) if the general measument changes....great.
4) if you want to assert that computers only mimic or copy intelligence ok.
5) fwiw I dont think computers pass a general test of measurement, today.
6) I sidestep the entire question of can a computer be self aware, alive or be conscious.
7) vast advances in the research and understanding of how the brain works will be needed to take the next step to an intelligent computer.

As I stated I have talked with many computer sci profs who all ..all agree that there is nothing in the laws of science that says a computer cannot be intelligent. However there is wide disagreement on how long that will take.

If it takes only 17 years lets discuss it ...if 17,000 years or longer perhaps not.

I'll try to answer your points in order:
1) I'll have to leave this one to you. When I search for a means of measurement I keep going back to Turing's "indistinguishable from human intelligence", and that didn't work out well.
2) Not sure I will be around to do this.
3)Noted
4)At present yes.
5)Agreed
6)Noted
7) Agreed

As regards how long, fear very long to never!

Sorry this isn't more helpful.

[mike777]

Scarabin, on 2013-February-09, 01:48, said:

I'll try to answer your points in order:
1) I'll have to leave this one to you. When I search for a means of measurement I keep going back to Turing's "indistinguishable from human intelligence", and that didn't work out well.
2) Not sure I will be around to do this.
3)Noted
4)At present yes.
5)Agreed
6)Noted
7) Agreed

As regards how long, fear very long to never!

Sorry this isn't more helpful.

you seem to want to not discuss point one.....


agree there is a agreement for and to measure.....all else in nonsense


If we cannot agree on how to measure ...rest is silly.


You keep claiming no agreement =with nothing so silly to discuss.

[onoway]

mike777, on 2013-February-08, 22:21, said:

1) lets for starters agree the definition of intelligence and how to test and measure it as generally accepted.


If it takes only 17 years lets discuss it ...if 17,000 years or longer perhaps not.

So where do you want to go with this?

[barmar]

dwar0123, on 2013-February-08, 11:57, said:

How does that logic hold up if computers can do everything(intellectually) that humans can do?

I was referring to what computers can actually do (i.e. what we've programmed them to do so far), not what they can theoretically do.