We are no different than computers

[Alex K]

Is the Dostoevsky test like the fictional Voight Kampff test?

[Quatermass]

We do differ from computers and robots in various ways because we have an associative memory - which is the main difference.

If you program a computer to do a certain thing, and program a computer to do the same thing, both computers will do that thing regardless of the fact the other is doing it.

If you program a robot to pick up a box and move it to another place, but remove the box, it'll still walk over to where the box was and pick up the box.

Now if you tell a human being to pick up a box, and remove the box, it wonders where the box went. If you assign it a task, and assign another human a task, it asks 'Why are we both doing the same task?'

This is all because even before were programmed (conditioned) by environment, we still build up various associations straight from birth which inform the way we react to our programming when it eventually starts later.


Also, computers can do much more than humans because they can process more information faster. A computer can process trillions of bits of information a second where as the human brain cannot (it only has a raw clock speed of about 20 billion bits per second).

You also couldn't merge computers with human brains properly using implants and stuff. The reason for this is that electricity flows through the brain to make memories or associations at a fairly slow rate, while in a circuit, electricity flows at nearly the speed of light. If you try to speed up the flow of electricity through the brain using a chip, nothing will happen (except maybe you cook the brain).

[Alex K]

We do differ from computers and robots in various ways because we have an associative memory - which is the main difference.

See above. You can in principle program a computer to have that.

[The Grand Nudger]

Guys, read about artificial neural networks. We are not talking about writing an algorithm that acts in some limited fixed ways here, as the colloquial use of the word "programming" would suggest.

This is true enough.  In fact, in a complex ANN, we STILL won't know how complex ideas form. Much as with the brain (maybe more so), we will only be able to wave toward our ANN and say, "Something happened in there. . ."
@OP
My question is this. Is it safe therefore to ASSUME that something which behaves like an emotional human therefore has an actual mind and actual feelings? How do you tell the difference between an actually sentient being and one that might seem to be sentient, but really isn't? Should robots that can pass the Turing test be given human rights? Should disabling one count as murder?

As Jor asked, how do you tell now?  The point of asking that question is not to cast doubt on whether or not you have a mind..but to help you to realize that you already employ a method in making the determination....and the method you use now, to decide that I, for example, have a mind...doesn't actually help you to determine whether or not -I- am "faking it".

My question is this. Is it safe therefore to ASSUME that something which behaves like an emotional human therefore has an actual mind and actual feelings? How do you tell the difference between an actually sentient being and one that might seem to be sentient, but really isn't? Should robots that can pass the Turing test be given human rights? Should disabling one count as murder?

The Turing test, no. Allow me to introduce the Dostoevsky test. When an entire dictionary is uploaded into a computer's language program, and pressed to take some time to express its "emotions" and "thoughts," if it can return with a work that approaches The Brothers Karamazov, perhaps writing notes about the process and how it came up with the ideas or subplots on the side, it will have sufficiently convinced me. Shit, I'd be happy if it produced something akin to Genesis 1 or even Dr. Seuss. Come to think of it, has anyone tried to write a book via a computer that formulates from its own software meaningful and/or artful syntax?

---

Well, this is interesting:
http://singularityhub.com/2014/11/09/com...read-them/

......and yet many creatures exist which we claim "mind" for (with tons of provisions..granted) - that could not pass the test.  You're expecting a machine to be a human being in order to -prove- that it has a mind...but doing so would more accurately speak to whether or not the machine was human, not whether or not it had "mind".  Eh?
We simply don't know how human beings "do mind".  We have the beginning of an inkling, that's it.  Comp mind offers an n as to how mind might be achieved by asking how things like comparison -are- achieved....in machine implementation.  Perhaps our minds do comparison differently - here are some examples of how it -can be- done. Explaining the unknown by reference to the known.  No one actually expects to find an analog circuit in your skull...........nor do we expect to be able to create "human" machines......that's not the suggestion......and it's probably not the most meaningful point of reference to begin with.
-@Quatar, swarming algorithms actually do posess the ability to assess workload in the manner you described as being a "human difference". Most notable examples are ant sims. We know a little bit about how actual ants "consider" their labor allocation as well - they don't. An effect is achieved which seems, to us, to be labor distruibution...it is distribution-of-labor- but nary a single ant ever gives a moments consideration to the matter. Amazing stuff, really - very efficient, very risk adverse.

[Thumpalumpacus]

How do you do that now? How do you know a psychopath or an autistic person has a mind?  How do you know you or I have a mind?

Personal reporting. Would a mindless thing have the creativity to lie and assert that it had a mind?

[The Grand Nudger]

I can point you to a handful of programs that do that right now (and will even engage in a conversation on the matter with you).  Creativity isn't a requirement for a lie, nor is mind.....machines can claim anything that you can.

[Thumpalumpacus]

I can point you to a handful of programs that do that right now (and will even engage in a conversation on the matter with you).  Creativity isn't a requirement for a lie, nor is mind.....machines can claim anything that you can.

Can it do it without software instructing it to do so? Does it have the self-volition to be deceptive?

And I'd disagree with your claim that creativity is not a requirement for lying.  The very act of lying is, in essence, the creation of another, fictional, reality.

[The Grand Nudger]

Will you keep adding modifiers if it turns out that machines can do these things - as you just did?  

Disagree all you like, but I think that lying can be described a little more simply than the creation of some alternate reality Parker.  Like...claiming something you both know to be untrue..which is so blissfully easy to do in a machine setting it pains me to tell you that all that is required is the negation of a true statement, voila...a lie has been produced.  All of this is ignoring how uncreative our every day human lies are to begin with. :wink:
We build negation into the alu's that control every cpu of every device with a computer circuit. -All- computers are capable of lying, we simply program them to do other, more useful things. Find me a useful lie and I bet you'll find machines generating those as well. While I don't think that negation encapsulates every idea you and I may have attached with lying, it -completely describes- the operative effect of a lie- by everyday use of the simplest possible gate.

[Exian]

But when a human (and some monkeys) lies, they employ theory of mind, empathy, creativity, and a desire gain a benefit. A computer will lie when it's told.

[Alex K]

But when a human (and some monkeys) lies, they employ theory of mind, empathy, creativity, and a desire gain a benefit. A computer will lie when it's told.

Not when you simulate an AI  on the computer which does these things to some extent. But you can have "lies" more easily - just write a program that gets some input and only relays true information if it serves some goal, and otherwise relays false information.