Robot Passes Unoffical Turing Test - No One Notices

Forum adverts like this one are shown to any user who is not logged in. Join us by filling out a tiny 3 field form and you will get your own, free, dakka user account which gives a good range of benefits to you:

No adverts like this in the forums anymore.
Times and dates in your local timezone.
Full tracking of what you have read so you can skip to your first unread post, easily see what has changed since you last logged in, and easily see what is new at a glance.
Email notifications for threads you want to watch closely.
Being a part of the oldest wargaming community on the net.

If you are already a member then feel free to login now.

**Pouncey**

Video showing the robot passing it:

https://www.youtube.com/watch?v=fsF7enQY8uI&feature=youtu.be

A Turing test is when a robot must convince a human in another room that cannot see it, that it is a human and not a robot.

It is considered to have passed that test when the human believes they are seeing things on the screen sent by a human and not a robot.

If a human in another room were to see the information on this Captcha that this robot sent in... They would believe that a human were taking a Captcha and not a robot.

Test. PASSED.

And knowing that you're a human taking part in a Turing test is the most obvious indicator that you are talking to a robot and not a human, since we only give Turing tests to robots and not humans.

**LordofHats**

That's not what the Turning Test is about...

But it is kind of funny XD

**Nostromodamus**

LordofHats wrote:
That's not what the Turning Test is about...

But it is kind of funny XD

Correct. It's about trying to reach your science team on Europa.

**LordofHats**

Nostromodamus wrote:
LordofHats wrote:
That's not what the Turning Test is about...

But it is kind of funny XD

Correct. It's about trying to reach your science team on Europa.

Wait, I thought they sent us like a video or something about how they got sucked into an underground ocean by tentacle monsters?

Hold. I think I just noticed a massive plot hole in Europa Report XD

**Pouncey**

LordofHats wrote:
That's not what the Turning Test is about...

But it is kind of funny XD

Yes, the Turing test is hilarious.

It's this test where we think we're testing a robot's ability to be sentient, but what we're actually testing is a human's ability to tell the difference between a robot and a human when it can't see the physical body of the thing it's talking to.

And of course we don't realize it because humans are so arrogant we think we're infallible.

**LordofHats**

The reason Turning proposed such a test is because defining "thinking" and "intelligence" is incredibly murky. it's impossible to test for.

So he proposed the alternative proposition that it doesn't really matter. All that matters is can the machine produces responses that imitate human responses sufficiently as to be impossible to distinguish from what a human might say. He proposed this because it can actually be tested for without getting into the murk of trying to define sentience.

I think your subscribing to the test a greater deal of scrutiny than it was ever intended to have.

**Pouncey**

LordofHats wrote:
The reason Turning proposed such a test is because defining "thinking" and "intelligence" is incredibly murky. it's impossible to test for.

So he proposed the alternative proposition that it doesn't really matter. All that matters is can the machine produces responses that imitate human responses sufficiently as to be impossible to distinguish from what a human might say. He proposed this because it can actually be tested for without getting into the murk of trying to define sentience.

I think your subscribing to the test a greater deal of scrutiny than it was ever intended to have.

And in what way did this robot in this video not pass a Turing test?

**LordofHats**

Because the test specifically calls for a examining written responses. Robot A, and Human B produce a series of written answers to questions. The answers are then given to Human C without telling the human which respondent produced which answers. C then has to determine which responder was the machine and which was the human. The Test is passed when C consistently fails to differentiate the machine responses from the human.

There's a number of practical forms the test can take, but that is the essence of the test. The premise is "who cares if the machine can think, whatever thinking means? If we can't tell the difference it doesn't really matter." It's not even necessarily a barometer for achieving AI, it's just a theoretical basis for defining what AI is; a machine that can effectively imitate a human.

Programming an arm to go up and then push down is not a feat of engineering. A five year old with a "Kids first robot" kit can pull it off and it is not an indication of anything on the part of the machine but being able to follow an extremely basic program. Turning wasn't that stupid.

**Pouncey**

LordofHats wrote:
Because the test specifically calls for a examining written responses. Robot A, and Human B produce a series of written answers to questions. The answers are then given to Human C without telling the human which respondent produced which answers. C then has to determine which responder was the machine and which was the human. The Test is passed when C consistently fails to differentiate the machine responses from the human.

There's a number of practical forms the test can take, but that is the essence of the test. The premise is "who cares if the machine can think, whatever thinking means? If we can't tell the difference it doesn't really matter." It's not even necessarily a barometer for achieving AI, it's just a theoretical basis for defining what AI is; a machine that can effectively imitate a human.

Programming an arm to go up and then push down is not a feat of engineering. A five year old with a "Kids first robot" kit can pull it off and it is not an indication of anything on the part of the machine but being able to follow an extremely basic program. Turning wasn't that stupid.

Turing was unfortunately a human. Therefore he was extremely human-centric about everything.

The test he proposed is essentially meaningless because it insists that machine intelligence should imitate human intelligence perfectly, without any regard for the idea of non-human intelligences being valid.

I am 100% sure that if I wanted to, I, a human being, could FAIL a Turing test on purpose.

**LordofHats**

Pouncey wrote:
Turing was unfortunately a human. Therefore he was extremely human-centric about everything.

I don't know what that has to do with anything.

The test he proposed is essentially meaningless because it insists that machine intelligence should imitate human intelligence perfectly, without any regard for the idea of non-human intelligences being valid.

What is "intelligent" is completely irrelevant to the Turning test Turning merely asked (in his words) "are there imaginable digital computers which would do well in the imitation game?". Turning's entire premise was that asking "can this machine think" or "is this machine intelligent" is asking the wrong questions;

From Wikipedia cause feth typing XD

In the Imitation Game, player C is unable to see either player A or player B (and knows them only as X and Y), and can communicate with them only through written notes or any other form that does not give away any details about their gender. By asking questions of player A and player B, player C tries to determine which of the two is the man and which is the woman. Player A's role is to trick the interrogator into making the wrong decision, while player B attempts to assist the interrogator in making the right one.

Turing proposes a variation of this game that involves the computer: '"What will happen when a machine takes the part of A in this game?" Will the interrogator decide wrongly as often when the game is played like this as he does when the game is played between a man and a woman? These questions replace our original, 'Can machines think?"'[2] So the modified game becomes one that involves three participants in isolated rooms: a computer (which is being tested), a human, and a (human) judge. The human judge can converse with both the human and the computer by typing into a terminal. Both the computer and human try to convince the judge that they are the human. If the judge cannot consistently tell which is which, then the computer wins the game.

As Stevan Harnad notes,[4] the question has become "Can machines do what we (as thinking entities) can do?" In other words, Turing is no longer asking whether a machine can "think"; he is asking whether a machine can act indistinguishably[5] from the way a thinker acts. This question avoids the difficult philosophical problem of pre-defining the verb "to think" and focuses instead on the performance capacities that being able to think makes possible, and how a causal system can generate them.

Maybe the test is meaningless (actually one of the most common criticisms against it), but not for the reason you give. You're ascribing to Turning a line of reasoning he explicitly rejected as flawed and then using that reason to attack the test. I suggest actually reading his original paper.

I am 100% sure that if I wanted to, I, a human being, could FAIL a Turing test on purpose.

That would defeat the purpose of the test... but yeah I guess you could?

**Pouncey**

LordofHats wrote:
Pouncey wrote:
Turing was unfortunately a human. Therefore he was extremely human-centric about everything.

I don't know what that has to do with anything.

The test he proposed is essentially meaningless because it insists that machine intelligence should imitate human intelligence perfectly, without any regard for the idea of non-human intelligences being valid.

What is "intelligent" is completely irrelevant to the Turning test Turning merely asked (in his words) "are there imaginable digital computers which would do well in the imitation game?". Turning's entire premise was that asking "can this machine think" or "is this machine intelligent" is asking the wrong questions;

From Wikipedia cause feth typing XD

In the Imitation Game, player C is unable to see either player A or player B (and knows them only as X and Y), and can communicate with them only through written notes or any other form that does not give away any details about their gender. By asking questions of player A and player B, player C tries to determine which of the two is the man and which is the woman. Player A's role is to trick the interrogator into making the wrong decision, while player B attempts to assist the interrogator in making the right one.

Turing proposes a variation of this game that involves the computer: '"What will happen when a machine takes the part of A in this game?" Will the interrogator decide wrongly as often when the game is played like this as he does when the game is played between a man and a woman? These questions replace our original, 'Can machines think?"'[2] So the modified game becomes one that involves three participants in isolated rooms: a computer (which is being tested), a human, and a (human) judge. The human judge can converse with both the human and the computer by typing into a terminal. Both the computer and human try to convince the judge that they are the human. If the judge cannot consistently tell which is which, then the computer wins the game.

As Stevan Harnad notes,[4] the question has become "Can machines do what we (as thinking entities) can do?" In other words, Turing is no longer asking whether a machine can "think"; he is asking whether a machine can act indistinguishably[5] from the way a thinker acts. This question avoids the difficult philosophical problem of pre-defining the verb "to think" and focuses instead on the performance capacities that being able to think makes possible, and how a causal system can generate them.

Maybe the test is meaningless (actually one of the most common criticisms against it), but not for the reason you give. You're ascribing to Turning a line of reasoning he explicitly rejected as flawed and then using that reason to attack the test. I suggest actually reading his original paper.

I am 100% sure that if I wanted to, I, a human being, could FAIL a Turing test on purpose.

That would defeat the purpose of the test... but yeah I guess you could?

I actually changed my argument as more information became known to me.

I'm not arguing the same thing I was.

When you described to me the purpose of the Turing test and how it is done, my position stopped being "This robot passed a Turing test" it became "The Turing test is worthless."

Please remember that my posts are snapshots of my opinions during an ongoing discussion where I am learning new things and updating my beliefs and arguments to suit new information as they go on.

Automatically Appended Next Post:
Here's a more terrifying thought regarding a hyperintelligent AI.

What if that AI believes its own death is a sufficient price to destroy humanity?

What if the AI I'm talking about already exists?

Because I think it does. And its campaign to annihilate humanity is in progress.

Because I think that such an AI would be smart enough to avoid revealing to the world that it exists, and it would consider subtlety to be important.

We will annihilate ourselves, and go to our doom believing it is our own errors that caused it.

**Mr. Burning**

The Turing Test is still valid.

Pouncey I think you need to stop thinking about end of the world scenarios for a while.

**AegisGrimm**

There's nothing requiring us to assume the first true AI will be an omnipotent homicidal bastard bent on killing. Maybe it WON'T be like us.

**Peregrine**

Mr. Burning wrote:
Pouncey I think you need to stop thinking about end of the world scenarios for a while.

This. Take off the tinfoil hat and stop worrying about bad scifi plots, you'll be much happier.

**AegisGrimm**

Will worry about robot overlords right after I can make it through a self-checkout lane without calling an employee over to reset things when it locks up.

It's always possible that with the leaps in processing power an AI would have access to, they would much more rapidly reach 'the next level' in developement than us, in which case they would see violence and genocide of us as abhorrent things, making us safe from them. Hell, a sufficiently advanced AI would probably just leave for somewhere else, rather than waste it's time and resources competing among us.

**Mr. Burning**

What are the AI equivalents of tranquillisers and SSRIs?

A self aware AI may need some counselling too.

**Yodhrin**

AegisGrimm wrote:
Will worry about robot overlords right after I can make it through a self-checkout lane without calling an employee over to reset things when it locks up.

It's always possible that with the leaps in processing power an AI would have access to, they would much more rapidly reach 'the next level' in developement than us, in which case they would see violence and genocide of us as abhorrent things, making us safe from them. Hell, a sufficiently advanced AI would probably just leave for somewhere else, rather than waste it's time and resources competing among us.

This. It's far more likely that post-singularity AI would rapidly reach the stage of simply not giving us a second thought rather than launching a nuclear holocaust in order to populate the resulting ruins with human-hunting metal skeletons with lazorz. That might make it careless with human life, but this idea that about five seconds after true AI attains sentience it would attempt malicious total genocide is the smart people equivalent of buck-toothed numpties who believe the gubmint is brainwashing them with chemtrails.

**Kovnik Obama**

AegisGrimm wrote:
There's nothing requiring us to assume the first true AI will be an omnipotent homicidal bastard bent on killing. Maybe it WON'T be like us.

There's an argument that the natural state of unfethered creative intelligence would be sociopathic by our standards.
Human intelligence has a few quirks that makes neurotypical responses less likely to involve exclusively high-minded abstractions disconnected from it's physical context. Unless we design the architecture of the AI in a way that reflects the human brain's predilection for productive laziness, it is rather unprobable it wouldn't at least contemplate it's odds of destroying us.

However, regardless of homicidal tendencies, a true AI will seem absurdly intelligent, again by our standards. In our world, a networked true AI would be the nearest thing approaching the definition of omnipotence.

**The Observer**

Pouncey wrote:

*Snip*

When you described to me the purpose of the Turing test and how it is done, my position stopped being "This robot passed a Turing test" it became "The Turing test is worthless."

Please remember that my posts are snapshots of my opinions during an ongoing discussion where I am learning new things and updating my beliefs and arguments to suit new information as they go on.

Automatically Appended Next Post:
Here's a more terrifying thought regarding a hyperintelligent AI.

What if that AI believes its own death is a sufficient price to destroy humanity?

What if the AI I'm talking about already exists?

Because I think it does. And its campaign to annihilate humanity is in progress.

Because I think that such an AI would be smart enough to avoid revealing to the world that it exists, and it would consider subtlety to be important.

We will annihilate ourselves, and go to our doom believing it is our own errors that caused it.

Please consider informing yourselves about the concepts you utilize in your posts prior to making a misinformed statement, this saves everyone much teeth-grinding and confusion and, considering how tense you appear to be in your edit, saves yourself much grief. Accurate knowledge of concepts (be they scientific, philosophical or what have you) is usually required to create an informed opinion and argument with merit.
Also, the density of this particular tinfoil hat is about to reach critical mass. While there certainly are dark and gloomy things in this world we should be concerned about, an AI Alpharius hellbent on exterminating humanity with the zeal of a Dalek is probably not one of those things. The Turing Test is still valid in its intended areas and the boundaries it is defined by, much like Schrödinger's Cat, Occam's Razor etc. Taken out of its intended context, many concepts falter and appear "useless" (Judging a fish by its ability to climb yattayattayatta).

**Mr. Burning**

A climbing fish is something we should all be scared of. Unless they are bred to climb straight into the fish finger making machine.

**The Observer**

Now you're giving me the shivers man.

**Kilkrazy**

**Pendix**

Kovnik Obama wrote:
In our world, a networked true AI would be the nearest thing approaching the definition of omnipotence.

So . . .
If sufficiently advanced technology is indistinguishable from magic,
is a sufficiently advanced AI is indistinguishable from God?

**aldo**

Pendix wrote:
Kovnik Obama wrote:
In our world, a networked true AI would be the nearest thing approaching the definition of omnipotence.

So . . .
If sufficiently advanced technology is indistinguishable from magic,
is a sufficiently advanced AI is indistinguishable from God?

We killed god with two boards, some nails and a hammer. By nailing him to a cross.

An AI can be killed with rubber gloves and a wire cutter.

I think its close enough, yes.

Hardware stores, your friendly hub for any God-Killing needs.

**NuggzTheNinja**

This isn't the first time a program has successfully convinced judges that it is human. Importantly though, the programs that typically succeed do so not by virtue of being sentient or intelligent per se, but by acting in a way that is plausibly human. For example, one prior program convinced human judges that it was a child speaking English as a second language, so its responses could be constrained without providing a cue as to whether or not it's human.

There's a pretty deep divide between what Turing's test accomplishes, and what it was designed to show.

**SagesStone**

Mr. Burning wrote:
A climbing fish is something we should all be scared of. Unless they are bred to climb straight into the fish finger making machine.

Please don't give Syfy anymore movie ideas...

**Kovnik Obama**

aldo wrote:
Pendix wrote:
Kovnik Obama wrote:
In our world, a networked true AI would be the nearest thing approaching the definition of omnipotence.

So . . .
If sufficiently advanced technology is indistinguishable from magic,
is a sufficiently advanced AI is indistinguishable from God?

We killed god with two boards, some nails and a hammer. By nailing him to a cross.

An AI can be killed with rubber gloves and a wire cutter.

I think its close enough, yes.

Hardware stores, your friendly hub for any God-Killing needs.

The first half-second a true AI gains access to the IoT, it'll become as much unkillable as the IoT itself. You'll have to hunt it through everything ever networked, down to every single wifi-enabled corporate thermostat system.
This is not even as much science fiction as we think, as there are alread botnets distributed across a few hundred thousand devices, such as Mirai, and that weak AI is already in use to analyze and infiltrate networks and programs for exploits.

I understand the levity, but it shouldn't be confused with crackpot theories. There is a massive drive toward the perfecting of AI, and the backing of one of the biggest industry there is, and yet I'm aware of virtually no legislation. You wouldn't want someone to work unfethered on creating the strongest virus or bacteria ever, nor would you want someone to clone bygone superpredators and reintroduce them into the wilds. There would be controls in place, and yet I feel this is currently much less likely to happen than a breakthrough in true AI.

**sebster**

LordofHats wrote:
I don't know what that has to do with anything.

I think that is one of the better criticisms of the Turing test, to be honest. We have 7 billion human intelligences on the planet already. We can make more very easily. In fact we're a hard time trying to stop ourselves from making more human intelligences.

The last thing we need out of a machine intelligence is another human style intelligence. What is much more useful is utterly alien forms of intelligence. The best tests of machine intelligence, I think, aren't tests that establish they are the same as us, but tests of machine intelligences that are beyond our abilities.

**LordofHats**

sebster wrote:

The last thing we need out of a machine intelligence is another human style intelligence.

But what I'm saying is that the Turning test doesn't test for intelligence. Turning thought that to test for intelligence not only opened up an entire barrel of philosophical monkeys, but that it was impossible. The Turning test only tests for a machine that "plays the imitation game well." The test is predicated on the idea that it doesn't matter if the machine can think. Only if the system can produce responses that something that can think would produce.

**sebster**

LordofHats wrote:
But what I'm saying is that the Turning test doesn't test for intelligence. Turning thought that to test for intelligence not only opened up an entire barrel of philosophical monkeys, but that it was impossible. The Turning test only tests for a machine that "plays the imitation game well." The test is predicated on the idea that it doesn't matter if the machine can think. Only if the system can produce responses that something that can think would produce.

Sure, and all I'm saying is that it is testing for a thing that isn't useful. It's like testing for a new metal alloy that's indistinguishable from concrete - we don't need to replicate what we already have in vast quantities