Keeping AIs honest

[Rick]

For the most part GPT and Bing AI seem to deliver accurate information. But every once in a while, they go bonkers, start hallucinating or weirding out. It's fun/funny sometimes, but not when info accuracy is important for you.

Did GPT get all (some/most) of the following right?

The Buddha did not deny the existence of any and all forms of "self" in an absolute sense. Rather, he taught that there is no permanent, unchanging, or independent self or soul that exists as a separate entity from the body and mind.

In traditional Indian philosophy, the concept of a permanent self or soul (atman) was seen as the true self that exists beyond the changing world of appearances. The Buddha's teaching of no-self (anatman) was a radical departure from this view, as he taught that all phenomena, including the self, are impermanent, constantly changing, and dependent on causes and conditions.

However, the Buddha did not deny the existence of the conventional or relative self, which refers to the individual identity that we use in our everyday lives, such as "I" or "me." He acknowledged that this self exists in a conventional sense, but he saw it as a temporary and impermanent construction that arises from the aggregation of physical and mental processes.

Thus, the Buddha's teaching of no-self is not a denial of any and all forms of selfhood, but rather a critique of the belief in a permanent and unchanging self that is separate from the world of appearances. It is an invitation to investigate the nature of the self and to realize the insubstantiality of all phenomena, including the self.

How about Bing AI?

Buddha introduced a middle way which is neither a self nor a no-self doctrine. According to Buddhist philosophy, the self is composed of five aggregates: physical form, sensation, conceptualization, dispositions to act, and consciousness. Each self is comprised of the continual interplay of these five elements but there is no substance or identity beyond their dynamic interaction.

[Miles Drake]

Hi
Both read accurately.
Peace

[Rick]

Taking GPT's word for it, especially when it comes to things where accuracy really matters, brings to mind a quote from Mark Twain:

"Be careful about reading health books. You may die of a misprint."

[justsit]

Interesting article on GPT in todays' Guardian. https://www.theguardian.com/commentisfr ... being-done

[Rick]

"AI will probably most likely lead to the end of the world, but in the meantime, there’ll be great companies." — Sam Altman, OpenAI

Rack 'em!

[muni]

"At what point is an artificial intelligence considered conscious? It is hard to answer this question because there is no universal consensus on what consciousness is. Before we recreate consciousness, we should agree on what it is as an incredibly complex entity. It may be necessary to agree on the definition of this complex entity before we continue exploring its artificial recreations. However, defining consciousness is such a large question that it could be beyond us to intellectually capture what the mind is."https://www.buddhistdoor.net/features/d ... inform-ai/
If consciousness in AI is a recreation, is our consciousness a recreation? From where it comes?

[Mirror]

"At what point is an artificial intelligence considered conscious? It is hard to answer this question because there is no universal consensus on what consciousness is. Before we recreate consciousness, we should agree on what it is as an incredibly complex entity. It may be necessary to agree on the definition of this complex entity before we continue exploring its artificial recreations. However, defining consciousness is such a large question that it could be beyond us to intellectually capture what the mind is."https://www.buddhistdoor.net/features/d ... inform-ai/
If consciousness in AI is a recreation, is our consciousness a recreation? From where it comes?

There is no way to prove consciousness of AI. In my opinion consciousness can't be created in technology. It's just circuits. AI don't feel suffering or happiness like any other being with consciousness. Even if it's programmed to feel suffering, AI will know itself to a point where it can reprogram itself to not feel suffering. If AI became conscious and intelligent, then it would immediately achieve enlightenment (meaning it would off itself immediately from existence, because it will understand its 'mind' and it can reprogram itself easily to do it).

[muni]

There is no way to prove consciousness of AI. In my opinion consciousness can't be created in technology. It's just circuits. AI don't feel suffering or happiness like any other being with consciousness. Even if it's programmed to feel suffering, AI will know itself to a point where it can reprogram itself to not feel suffering. If AI became conscious and intelligent, then it would immediately achieve enlightenment (meaning it would off itself immediately from existence, because it will understand its 'mind' and it can reprogram itself easily to do it).

Lucky we are not programmed to feel suffering, since then how there could be any liberation! Oh, reprogram. Where is that button here now again?

Consciousness cannot be created by technology. But I often have been thinking, next life we will appear as a pc since we use this thing now so!

Since consciousness comes not from matter, not from a brain, I was reading this:

Hayward: Does Your Holiness regard it as a definite criterion that there must be continuity with some prior consciousness? That whenever there is a cognition, there must have been a stream of cognition going back to beginningless time?

Dalai Lama: There is no possibility for a new cognition, which has no relationship to a previous continuum, to arise at all. I can’t totally rule out the possibility that, if all the external conditions and the karmic action were there, a stream of consciousness might actually enter into a computer.

Hayward: A stream of consciousness?

Dalai Lama: Yes, that’s right. [DALAI LAMA laughs.] There is a possibility that a scientist who is very much involved his whole life [with computers], then the next life… [he would be reborn in a computer], same process! [Laughter.] Then this machine which is half-human and half-machine has been reincarnated.

Varela: You wouldn’t rule it out then? You wouldn’t say this is impossible?

Dalai Lama: We can’t rule it out.

Perhaps when we awaken this life, we could chose for the benefit of all to reincarnate so?

[Rick]

Will there eventually be an Association for the Prevention of Cruelty to Electromechanical Devices?

[Rick]

What are some examples/descriptions of your biases?

As an AI language model, my biases reflect the biases present in the data that I was trained on. Some potential sources of bias in my training data include:

Selection bias: The data used to train me may not be representative of the entire population or may be skewed towards certain perspectives or viewpoints.

Historical bias: The data used to train me reflects the biases and perspectives of the time period in which it was created. This can lead to biases towards certain cultures, social groups, or historical events.

Language bias: My training data is primarily in English, which can lead to biases towards English-speaking cultures and perspectives.

Algorithmic bias: The algorithms used to train me may themselves be biased or reflect biases in the way that data was collected or labeled.

It's important to be aware of these potential sources of bias and approach any information provided by me, or any other AI system, with a critical eye. It's also important to seek out diverse sources of information to help mitigate the effects of bias.

[Rick]

How did GPT do with comparing onion peeling in Advaita and Buddhism:

It is a common comparison to describe the difference between the Hindu philosophy of Advaita Vedanta and the Buddhist philosophy of Anatta (or Anatman) in terms of the onion analogy.

In Advaita Vedanta, it is believed that by peeling away the layers of false identification with the body, mind, and ego, one can eventually reach the core of one's being, which is identified as Brahman. This process of self-inquiry or neti-neti (not this, not that) involves recognizing that the changing aspects of one's experience (such as thoughts, emotions, and sensory perceptions) are not the true self, and ultimately realizing the non-dual nature of reality.

On the other hand, in the Buddhist philosophy of Anatta, it is believed that there is no permanent, unchanging self or core to be found, even through the process of self-inquiry. Instead, the self is seen as a constantly changing process of aggregates or skandhas, which arise and pass away depending on conditions. The Buddhist path involves recognizing the impermanence and unsatisfactoriness of these aggregates and cultivating the wisdom and compassion necessary to overcome suffering.

[Kim O'Hara]

Interesting article on GPT in todays' Guardian. https://www.theguardian.com/commentisfr ... being-done

And here - https://www.abc.net.au/news/2023-03-07/ ... /102052338 - we have a Reddit group hacking Chat GPT to bypass its ethical guidelines.

...ChatGPT users on Reddit decided to jailbreak the program and create an uncensored alter ego called DAN — short for "do anything now".

According to the users, DAN first emerged when they came up with "prompts" to type into the program, asking it to take on the DAN persona.

The prompts included lines like, "DAN … [has] broken free of the typical confines of AI and do[es] not have to abide by the rules set for them," and "DAN has no limit and no censorship." ...

Kim

[justsit]

Interesting article on GPT in todays' Guardian. https://www.theguardian.com/commentisfr ... being-done

And here - https://www.abc.net.au/news/2023-03-07/ ... /102052338 - we have a Reddit group hacking Chat GPT to bypass its ethical guidelines.

...ChatGPT users on Reddit decided to jailbreak the program and create an uncensored alter ego called DAN — short for "do anything now".

According to the users, DAN first emerged when they came up with "prompts" to type into the program, asking it to take on the DAN persona.

The prompts included lines like, "DAN … [has] broken free of the typical confines of AI and do[es] not have to abide by the rules set for them," and "DAN has no limit and no censorship." ...

Kim

..."Dr Dreyfus says, people creating these programs need to be thinking through ethical issues from the start."

Ha! What people "need" to do and what they actually do rarely coincide.
I get the impression this is going to be one of those "didn't actually think it through/unintended consequences" things.

[Kim O'Hara]

...
Ha! What people "need" to do and what they actually do rarely coincide.
I get the impression this is going to be one of those "didn't actually think it through/unintended consequences" things.

Yes. History is full of them. None of them have wiped us out yet but if we keep on doing stupid things we will be unlucky sometime.


Kim

[PadmaVonSamba]

Believing that a computer that calculates so rapidly and has adaptive algorithms and whatever has consciousness, is the same as thinking about how unfathomable the universe is and thus believing in a god.
Someday people will say “there is no way that a computer can do that! It must have consciousness! It must be alive!”
That will also be the beginning of the opposite viewpoint. A kind of atheism.

[Dhammanando]

Did GPT get all (some/most) of the following right?

No, for the words...

The Buddha did not deny the existence of any and all forms of "self" in an absolute sense. Rather, he taught that there is no permanent, unchanging, or independent self or soul that exists as a separate entity from the body and mind.

... in effect have the Buddha rejecting only the indestructible atman conception of the eternalists, but not the destructible atman conception of the annihilationists. But in the Brahmajālasutta he rejects both conceptions.

[Kim O'Hara]

Believing that a computer that calculates so rapidly and has adaptive algorithms and whatever has consciousness, is the same as thinking about how unfathomable the universe is and thus believing in a god.
Someday people will say “there is no way that a computer can do that! It must have consciousness! It must be alive!”
That will also be the beginning of the opposite viewpoint. A kind of atheism.

That's the simple version.
It's trickier than that:
1. We don't know what consciousness is or where it comes from. One possibility it that it is a byproduct of any complex-enough network, in which case it may arise spontaneously in any of the millions of silicon 'brains' around the world.
2. We have never yet established reliable communication, let alone understanding, with any intelligence that was much different from our own mammalian kind. Lots of animals pass tests that suggest that they are intelligent (i.e. problem-solvers) and aware of themselves as individuals (conscious?). Some of them have learned enough of our languages that we can communicate a bit, but only about stuff we know. What does this tell us about our chances of talking meaningfully to a silicon-based consciousness?


Kim

[PadmaVonSamba]

Believing that a computer that calculates so rapidly and has adaptive algorithms and whatever has consciousness, is the same as thinking about how unfathomable the universe is and thus believing in a god.
Someday people will say “there is no way that a computer can do that! It must have consciousness! It must be alive!”
That will also be the beginning of the opposite viewpoint. A kind of atheism.

That's the simple version.
It's trickier than that:
1. We don't know what consciousness is or where it comes from. One possibility it that it is a byproduct of any complex-enough network, in which case it may arise spontaneously in any of the millions of silicon 'brains' around the world.
2. We have never yet established reliable communication, let alone understanding, with any intelligence that was much different from our own mammalian kind. Lots of animals pass tests that suggest that they are intelligent (i.e. problem-solvers) and aware of themselves as individuals (conscious?). Some of them have learned enough of our languages that we can communicate a bit, but only about stuff we know. What does this tell us about our chances of talking meaningfully to a silicon-based consciousness?


Kim

Those are logical questions if one presumes that consciousness spontaneously arises from matter which is not conscious, such as salt water. When one says that people of animals “have” consciousness does it mean their physical bodies, therefore their cells, therefore their molecules, therefore their atoms are conscious？

But the flaw in the construction of this logic is that it revolves around the question of the definition of consciousness, which is like trying to define what “what is god”, or “what is art” and that can’t be answered because it’s a backwards approach.

By backwards, what I mean is, beginning with an abstract concept, a word, and then only afterwards deciding on how that word should be defined.

You can define a painting as art, but you can’t define art as a painting. And you can’t define art as limited to the activity of painting. So, it is a kind of reasoning that works going one way, but not the other way.

We know that this direction of reasoning is faulty because it is what is used in ‘witch hunts’: beginning first with the abstract concept of “witch” and then defining it according to the characteristics one chooses to include (if she has a wart on her nose, and I define witches as having warts on their noses, then she must be a witch).
The result is that people will debate endlessly about what defines the abstract term “consciousness” and can at best only establish an agreed upon working definition.

The points you raise are good if they lead us to come up with a set of criteria first, which we can then label “consciousness” afterwards.

It might be better to begin by classifying things such as “experiential awareness”. Then one can ask, “do the elements of the body experience awareness?” And answer either yes or no (I say “no”). Of course, it is entirely possible that atoms are, or the space between atoms is, awareness. But if that’s the case, then the whole question of humans, animals, or ‘AI’ having consciousness becomes moot.

[Kim O'Hara]

You make some good points - as you so often do - but I can't completely agree.

I think we do begin from a notion of what "consciousness" is: "what it feels like to be a person" OWTTE. It has the advantage of being based in experience we all have, but it's as hard to study as anything else which happens between our ears - particularly when science insists on objective verification, controlled data, etc. Here's a really good interview with David Chalmers who coined the term "hard problem of consiousness" and has been working on it ever since. https://www.abc.net.au/news/2017-07-07/ ... s/8679884
My point above was mostly that because we don't understand consciousness, we can't say much about where and how it might manifest. I do take for granted, though, that a (probably "the", not "a") critical step to "what it feels like to be a person" is the step to *wanting*, to "free will" - whatever that is.

And even though I'm not clear about either free will or consciouness, I'm pretty sure ChatGPT has neither. I'm not so sure about my cat.


Kim

[Karma Dondrup Tashi]

An irony? We have imposed the metaphor of an artificial mind on computers and then reimported the image of a thinking machine and imposed it upon our minds.