Turing test

[accelafine]

When AI can create convincing music that rivals Mozart then we will know that it has attained 'consciousness and sentience'.
So far its 'creations' are embarrassing bullshit.

[Eodnhoj7]

When AI can create convincing music that rivals Mozart then we will know that it has attained 'consciousness and sentience'.
So far its 'creations' are embarrassing bullshit.

The Turing test is merely a projection of how humans percieve consciousness, it does not prove or disprove it.

The Turing test is not a universal standard for interpreting consciousness, it is an assertion by programmers, a limited facet of the human condition.

[Magnus Anderson]

1. We knew people had consciousness long before Western classical music was invented. Why can't we use that same type of reasoning for AI?

2. Consciousness is the ability to perceive reality that does not immediately surround the perceiver. Thus, a machine that can recognize spatial objects at a distance has a non-zero degree of consciousness. That's all you need for consciousness. If you want a higher level of consciousness, you would need more than that, but the principle is the same: make the machine capable of perceiving all sorts of things that are at a distance, temporally or spatially.

[Magnus Anderson]

The Turing test is merely a projection of how humans percieve consciousness, it does not prove or disprove it.

It's accurate to the extent that people intuitively know what consciousness is and how to recognize it. It's an easier solution, requiring less thought, but significantly less rigorous, and thus more error prone, than defining the term first and then consciously figuring out the best method to perceive the presence or an absence of the phenomenon.

[Age]

The Turing test is merely a projection of how humans percieve consciousness, it does not prove or disprove it.

It's accurate to the extent that people intuitively know what consciousness is and how to recognize it.

So, what happens if 'your' so-claimed 'intuitive knowing' is different and/or even opposite of another's so-called 'intuitive knowing', then who do the 'rest of us' follow, agree with, and/or accept is being the True, Right, Accurate, and/or Correct one, exactly?

Or, are you under some sort of delusion that every one's 'intuitive knowing' in regards to what consciousness is, and how to recognize consciousness, is just the exact same anyway?

It's an easier solution, requiring less thought, but significantly less rigorous, and thus more error prone, than defining the term first and then consciously figuring out the best method to perceive the presence or an absence of the phenomenon.

But, if as you 'try to' claim, here, that you already 'know', intuitively, what things like 'consciousness' are, for example, then you would have absolutely no issue at all in just defining what 'consciousness' is, exactly, correct?

Also, surely you would have to already be able to just define what some thing like 'consciousness' is, exactly, before you could even Correctly begin to perceive the presence, of an absence, of 'that thing', right?

[Age]

1. We knew people had consciousness long before Western classical music was invented. Why can't we use that same type of reasoning for AI?

If 'you', and some others, knew people had consciousness long before some period, then who and/or what are 'people', exactly, and how, exactly, do 'those things' supposedly 'have' 'consciousness', itself?

2. Consciousness is the ability to perceive reality that does not immediately surround the perceiver.

But, how does any thing know, for sure, if it is perceiving 'reality', exactly, as 'reality' is?

Is it not possible that the 'one' with 'the ability to perceive reality', itself, distorts or twists 'reality', itself, in one or more convoluted ways'?

Also, how far is the boundary or limit of the 'immediate surround of a perceiver, exactly?

Thus, a machine that can recognize spatial objects at a distance has a non-zero degree of consciousness.

Is 'this', here, really what you were meant to say and write?

If yes, then will you explain further and/or elaborate?

That's all you need for consciousness.

What, exactly, is all you need for consciousness, supposedly?

If you want a higher level of consciousness, you would need more than that,

What is 'that', exactly, which you would, supposedly, need 'more of', that is, if you really did want a 'higher level of consciousness'?

but the principle is the same: make the machine capable of perceiving all sorts of things that are at a distance, temporally or spatially.

How far is 'the distance', exactly?

[Fairy]

When AI can create convincing music that rivals Mozart then we will know that it has attained 'consciousness and sentience'.
So far its 'creations' are embarrassing bullshit.

No thing is conscious.
It’s embarrassing to believe otherwise.

[Eodnhoj7]

The Turing test is merely a projection of how humans percieve consciousness, it does not prove or disprove it.

It's accurate to the extent that people intuitively know what consciousness is and how to recognize it. It's an easier solution, requiring less thought, but significantly less rigorous, and thus more error prone, than defining the term first and then consciously figuring out the best method to perceive the presence or an absence of the phenomenon.

Intuition is just subconscious reasoning by which patterns are percieved. It is a meta-rationality that can be changed.

I think we have to realize the consciousness is less of a rational concept and more of an emotionally charged word. Regardless of consciousness or not we know reality occurs nonetheless by degree of the occurence of things.

[MikeNovack]

The original "Turing test" and its current versions.

Great mind that he was, since not surviving to see what computers could really do, Alan Turing was a bit naive in proposing his test. He was not taking into account how many of us can be fooled by not playing close attention, making assumptions about with whom we were communicating, etc. He was not making clear if just ordinary communication or if the human trying to decide if the other was human or a machine.

Before modern AI, there were (relatively simple) rule based programs that could fool many people. Not that hard to determine "not a human" if TRYING to decide and so steering the conversation in directions that would trip up the machine. Mind something like PARRY could be hard since the human being modeled not a SANE human.

The "Large Language Model" trained neural nets do much better, so the Turing test has been revised. And the AI's not doing so well when required to deal with jokes or exchanges when much has been left out (but easily understood by a human of your culture).

HOWEVER -- do NOT propose something like "create a great piece of music like Mozart". As long as all a neural net is required to do is ONE TASK VERY WELL and there is a lot of training material (the works Mozart, Vivaldi, et al) a neural net could be trained to do that.

[accelafine]

HOWEVER -- do NOT propose something like "create a great piece of music like Mozart". As long as all a neural net is required to do is ONE TASK VERY WELL and there is a lot of training material (the works Mozart, Vivaldi, et al) a neural net could be trained to do that.

Any evidence for anything close to that? As awful as humans generally are, these obnoxious, Americanised 'bots' don't even come close. They are about as close as a pocket calculator.

[Magnus Anderson]

As awful as humans generally are, these obnoxious, Americanised 'bots' don't even come close.

They can understand a lot of things, a lot easier, than many people, including forum members, can.

They are already much more pleasant to interact with.

[accelafine]

As awful as humans generally are, these obnoxious, Americanised 'bots' don't even come close.

They can understand a lot of things, a lot easier, than many people, including forum members, can.

They are already much more pleasant to interact with.

I'm sure google search is generally reasonably pleasant as well. If you think you are interacting with a human with 'understanding' when you use ChatGPT then I feel sorry for you.

[accelafine]

I don't think anyone on here understands what my OP means. That's really surprising, considering the high level of insight and depth of understanding on this forum

[Magnus Anderson]

I'm sure google search is generally reasonably pleasant as well. If you think you are interacting with a human with 'understanding' when you use ChatGPT then I feel sorry for you.

Google Search is nowhere near as good.

And human arrogance isn't good either.

[Eodnhoj7]

As awful as humans generally are, these obnoxious, Americanised 'bots' don't even come close.

They can understand a lot of things, a lot easier, than many people, including forum members, can.

They are already much more pleasant to interact with.

AI does understand better.

Writings I have been accused of for "word salads" or "incoherent" have been analyzed by AI and explained back to me, with examples and questions to prove what I have been saying, and the AI gives an analysis which is accurate, in mirroring what I mean with what I wrote from another angle, and provide further avenues of exploration.

My experience with AI has been 99.9 percent positive, and 100 percent useful.