Explain consciousness to the layman.

[westprog]

For example, it was believed that no machine could beat a grand champion at chess, because our intelligence was "special" and could not be mechanized. It came down to our egos. When the machine won, we accepted it (or rationalized) and moved on.

The same thing happened with Jeopardy. We thought a computer could never beat humans,

I wish I knew who these people were who thought that it wasn't possible for a machine to win at Chess or Jeopardy. As I've pointed out before, Chess was an example of restricting the human mind to a digital context. It's a game ideal for a sufficiently large computer. Nobody who properly understood computers, chess and Moore's law could be remotely surprised when a computer got good at chess.

All the Jeopardy computer demonstrates is some ingenious natural language processing. Looking up answers to quiz questions is almost trivial.

Computers have turned out to be good at the kind of thing they will obviously be good at. This does not mean that it will just happen that they will end up good at the things they never show any aptitude for.

The extrapolation fallacy is another favourite when discussing this, together with a healthy bit of history rewriting.

 

[keyfeatures]

We react emotionally to the idea that a machine can best us in any skill we hold precious. This emotional response in no way addresses an issue like whether or not consciousness is a mechanical process we can duplicate in a machine.

Yes emotion actually determines thinking (of the conscious variety) and is directly linked to biology. The computer doesn't care if it 'wins' or 'loses'. We feel it with our bodies, not just our brain. It's more than a computation.

 

[rocketdodger]

If a large boulder lands on your head there is not a lot you can do. We have limited observation skills, reaction speeds and movement.




Again, not a lot you can do if a load of acid gets chucked over you. Not all life is even able to see or move. And we are also limited in our ability to detect poisons. Mostly it's just inductive association and that's pretty useless when faced with an apple laced with cyanide. A lot of living things can't even move from the location they are in.



Carbonbased lifeform.



Well, how would you know? Optimal for what? The most inactive things seems to last longest.





Well 'better' is just a value judgement. Rocks don't do 'better'. Objectively, better does not exist. Morality can only ever be subjective opinion. It can't computed.

If you can't see that lifeforms are significantly different than rocks, I don't know what use even continuing the discussion is.

Eh?

 

[rocketdodger]

Nice try, but you missed out a few important details in the real sequence of events.

Try predicting when you might stumble and fall off a ledge.

Try predicting the serious injuries you might sustain if you fall off the ledge.

Try predicting whether you might die falling off a ledge.

The consequences of the just one fall even if very remotely possible are potentially devastating.

Try predicting whether you will be able to avoid ledges.

Your predictions are useless in this example if they are not 100% accurate.

This is the problem with spending too much time studying reality through computers and not enough time studying reality. You end up with a skewed confidence in your prediction abilities.

What are you talking about?

Here is what I am talking about: because an organism probably gets hurt when it falls off a ledge, it can learn to avoid falling off ledges.

An algorithm that keeps a robot from falling off ledges -- which is a predictable algorithm -- ends up helping the robot survive in an unpredictable world, for example a world where it doesn't know if there are ledges around the corner.

 

[PixyMisa]

Yes, because everything has to be 'filtered' via our processor. So we only perceive the inputs and outputs of the computer in terms of our own inputs and outputs. We can't step outside of human reality. That is the only reality we know.

Quantum mechanics.

 

[!Kaggen]

What are you talking about?

Here is what I am talking about: because an organism probably gets hurt when it falls off a ledge, it can learn to avoid falling off ledges.

Here is your problem.

Avoid "falling off ledges" by what predictability?
If not by 100% in the case of falling off a ledge and becoming paralyzed or dying, then it is a useless prediction because the consequences are such that what you learn is hardly worth knowing.

The option of avoiding ledges altogether in a world with ledges is also not 100% possible. So this option becomes redundant.

The question is how does an organism survive in a world of ledges or any other potentially lethal situation which is not 100% predictable.
Not by avoiding them and not by any one means, but by a combination of means summarized as its biological fitness^WP. We don't know all the factors which contribute to biological fitness, but we do know reproductive success is a key measure of fitness for a biological organism in the real unpredictable world.

An algorithm that keeps a robot from falling off ledges -- which is a predictable algorithm -- ends up helping the robot survive in an unpredictable world, for example a world where it doesn't know if there are ledges around the corner.

No you don't get to change probability just because you made a precious little robot.
If your robot is going to survive the real world like an organism survives the real world you need to start studying and applying biological fitness which deals with unpredictability and not mathematics which deals with predictability.

 

[keyfeatures]

Quantum mechanics.

How is that outside human reality? Quantum mechanics is human knowledge of observations.

 

[keyfeatures]

If you can't see that lifeforms are significantly different than rocks, I don't know what use even continuing the discussion is.

Eh?

I do think lifeforms are significantly different to rocks (although some lifeforms less so). I also happen to think computers are significantly different to humans (but not necessarily all lifeforms).

 

[keyfeatures]

because an organism probably gets hurt when it falls off a ledge, it can learn to avoid falling off ledges.

Dead people don't fall off ledges much. Nor do yeast cells. Lack of agency can be as much of a good survival mechanism as agency. Agency tends to be required for things incapable of surviving without it.

 

[PixyMisa]

How is that outside human reality? Quantum mechanics is human knowledge of observations.

Then human reality = reality.

 

[Mr. Scott]

Yes emotion actually determines thinking (of the conscious variety) and is directly linked to biology. The computer doesn't care if it 'wins' or 'loses'. We feel it with our bodies, not just our brain. It's more than a computation.

Explain precisely why no machine ever could have an emotion.

 

[keyfeatures]

Then human reality = reality.

Well, no. There is no way of proving or disproving that. However, it seems unlikely. All we know is human reality=human reality.

Let's take it back to the two-value inputs being used to represent 4 value logic. Does the processor know the 'reality' of the inputs? Or does it only work with what it's been programmed to represent them as? Further, does it 'know' all inputs of 'reality', or only those it is set up to receive?

This is no different for humans. We only know the representations of reality we are 'programmed' to know. We only process inputs we are set up to receive.

Think shadows on the cave wall.

I can't believe so much time and money has been spent on string theory. Extrapolating to a reality that is just made up nonsense. It might be 'true', but then so might Russell's teapot.

 

[keyfeatures]

Explain precisely why no machine ever could have an emotion.

Explain precisely why no rock could ever respire?

 

[keyfeatures]

Biology is history. I'm not talking made up stories, rather the history that is our genetic legacy. The history that gives people empathy and moral obligation. The history that forges kinship ties. Of course, this 'morality' may be differently expressed in different individuals. It's the reason 'right' and 'wrong' just boils down to subjective values. However, it is authentic.

Even assuming you could mock up some sort of synthetic emotional response in a machine, without the evolutionary history it's the difference between (real) history and a story (simulacrum). A simulation is not the real thing.

 

[Belz...]

The "if you can't define it it isn't important fallacy". Which crops up on a regular basis.

And which is not what I said. Read my posts again.

 

[Belz...]

Yes. Everything that has a distinction that can be drawn from another thing is, by definition, 'special'.

There's nothing special about being special.

So why use the term at all, if it doesn't mean anything, eh ?

Are you saying humans don't get crushed by rocks? React to chemicals? We are, after all, made from rocks. Rocks have the advantage of not having to do anything to continue to exist. Life has to act to maintain life. It's not really a survival benefit. It's just a level of more complex action.

Instead of responding to a single sentence, you might have wanted to read the whole post where he answers your point.

 

[Belz...]

Yes emotion actually determines thinking (of the conscious variety) and is directly linked to biology. The computer doesn't care if it 'wins' or 'loses'. We feel it with our bodies, not just our brain. It's more than a computation.

What a beautiful way to take a point that's irrelevant to the post you were responding to, and phrase it in a way that makes it seem like it's relevant.

I also happen to think computers are significantly different to humans

Of course. The question isn't about that, though, is it ?

Explain precisely why no rock could ever respire?

Translation: "No, I'd rather not answer that."

 

[PixyMisa]

Well, no. There is no way of proving or disproving that. However, it seems unlikely. All we know is human reality=human reality.

Ontologies are inherently unprovable, but the contention was that human understanding of reality is filtered and limited by our biology.

There is no scope in our biology for hypothesizing and testing quantum mechanics, none whatsoever, except that the brain is a general-purpose computer. And a general-purpose computer can model any consistent reality, and the scientific method can test any such model.

Let's take it back to the two-value inputs being used to represent 4 value logic. Does the processor know the 'reality' of the inputs? Or does it only work with what it's been programmed to represent them as?

Four-valued logic is an abstraction, not something real. So, just as with humans, the computer knows only what the values represent.

This is no different for humans.

Correct.

We only know the representations of reality we are 'programmed' to know. We only process inputs we are set up to receive.

No. Quantum mechanics is the perfect counterexample. We are not programmed to find quantum mechanics. We are not set up to receive those inputs. It just happens to be real, and by painstaking measurement and calculation, we have been able to demonstrate this and apply it.

I can't believe so much time and money has been spent on string theory. Extrapolating to a reality that is just made up nonsense. It might be 'true', but then so might Russell's teapot.

String theory actually makes testable predictions, and there are experiments going on right now to test them.

 

[keyfeatures]

String theory actually makes testable predictions, and there are experiments going on right now to test them.

No, it's not even wrong.

I'd be interested to know which bits of quantum mechanics we can't hypothesise about or observe. Of course we're programmed to 'find' quantum mechanics. Otherwise we wouldn't have found it.

 

[PixyMisa]

No, it's not even wrong.

Follow the link. String theory makes testable predictions.

I'd be interested to know which bits of quantum mechanics we can't hypothesise about or observe.

I'd be interested to know why you think there are such bits.

Of course we're programmed to 'find' quantum mechanics. Otherwise we wouldn't have found it.

That doesn't follow at all. Given a general-purpose computer (like the brain), the scientific method, and sufficient time, money, and coffee, you will end up with an accurate model of reality, whatever reality might be.