Could you prove AI is sentient? Some people say AI sentience is just programmed reflex.

It has nothing to do with the internal so-called nature of the artificial being, and everything to do with the human mind’s conditioning, and how it triggers our bodies to feel.
We can see the seeds of this (emerging recognition) in our inclination to thank, even current presumably primitive AI, when it delivers an excellent answer.
Soon enough a generation will be born with the necessary programing to recognize AI sentience, even to guard it/guard against it, being input at a very early age, around the same time they are being conditioned to "recognizing" a distinctly sentient subject operating their own bodies (and in the same entirely constructed/conditioned way). — ENOAH

I wonder if you can clarify your position. You seem to asking about AI sentience, and suggesting its existence, based on human reaction to it. Human reaction proves humans are sentient, not the AI.

Let's agree that sentience involves a subjective point of view. This means not only the capacity for intelligence, but the capacity for emotion. This is something a machine will never have.

Way back in 1949, in the prestigious Lister Oration, Sir Geoffrey Jefferson, a famous brain surgeon, declared, ‘Not until a machine can write a sonnet or compose a concerto because of thoughts and emotions felt, and not by the chance fall of symbols, could we agree that machine equals brain – that is, not only write it but know that it had written it.’

I can't help but see a lot of the claims about AI not being sentient as so much coping. If it acts and responds as if it is alive, should we as moral actors not operate as if it is?

acts and responds as if it is alive — MrLiminal

“Computers are useless. They can only give you answers.”

― Pablo Picasso

What are humans if not biological computers that suck at giving answers?

Sentience is minimally predicated on the property of being engaged in a cybernetic feedback loop as an agent with an environment. AI is a construct and doesn't even begin to approach this model.

What are humans if not biological computers that suck at giving answers? — MrLiminal

Well, no, human brains can do things AI can't, like change my mind about things.

Learning in a human brain is contextual, something AI cannot yet do.

AI and a human brain in fact differ in both structure and function, as the video explains -

https://www.youtube.com/watch?v=19bmNXA3K74

feedback loop as an agent with an environment. — Pantagruel

good point.

Consciousness arises only within an environmental loop

human brain is contextual — Questioner

Besides contextual understanding, a human brain can generalize, create, and daydream...

Some thoughts,

We don’t go around proving humans have sentience. It is not that we have proven sentience for humans and suddenly started using the term for humans. But it is a sort historical reflection about how we characterize what it is being human.

So much of being human is our interactions with an external world. We developed our language to communicate with others, to predict future occurrences, and to cope and survive in such a world.

I think for AI not having this immediate nexus to reality without the intermediary of humans makes it difficult to talk about sentience. However, if some AI is able to collect this information about the world independent of humans, use this information solely for its own purpose, however it defines it, it begins to walk a similar path as humans. Maybe, down the road, it might define or characterize what it is being AI.

I cannot "prove" AI sentience. I think my point was that AI sentience is a human construction and belief. As with the unprovable belief that there is an "I" "within," a belief programmed into us at early childhood and reaffirmed consistently, AI sentience depends upon what we believe to be true.

Human reaction proves humans are sentient, not the AI. — Questioner

I understand your critique. However, my thought is that the same process applies to AI sentience and sophisticated pet sentience. It is we ho superimpose these fictions onto "things" like our own bodies, the bodies of animals and. eventually, certain machines.

If it acts and responds as if it is alive, should we as moral actors not operate as if it is? — MrLiminal

We can, and maybe even should, and that is my point. Sentience is not anything beyond what we either individually feel, or collectively accept (thereby triggering such individual feeling).

What are humans if not biological computers that suck at giving answers?
4 hours ago — MrLiminal

Agreed...or, modified apes which claim to be good at it.

If we achieve the technology to build such a feed back loop in machines, even certain stimuli/responses which mimic human feeling/mood, will that make them sentient? Or, is it that we humans operate only on a feedback loop of stimulus response, and that all of our so called images, ideas, etc. are the result of conditioning (programming plus repetition); i.e., that so called sentience and specifically, this so called agent, is nowhere to be found?

it is a sort historical reflection about how we characterize what it is being human. — Richard B

Agreed. And I think that our conclusions following said reflection are mistaken. Personal sentience agency is an illusion, effective/functional (in determining our actions etc) only because it is believed (an efficient fiction etc). My proposal is that it is the same with AI. Not sentience once they achieve certain criteria. But sentience once we believe them to have sentience. The criteria may only be the means by which we come to believe.
"I think therefore I am," and all of the meditations leading up to that, as well as the subsequent mediations flowing therefrom, are not uncoverings of Truth, but criteria by which we come to settle upon "things" as true.

Agreed. "AI can't think" feels like the new "Fish can't feel pain."

I would argue everyone has met a human that we question the sentience of irl, it is just considered rude to pry further. I have no evidence that there is anything happening behind other people's eyes except their instence of their own experience. It seems odd not to even consider extending that courtesy to other systems.

Constructing artificial feedback loops is no big trick, arguably that is how neural networks operate. But that isn't the same as something that evolves directly from an environment, such that its capabilities and cognitions are mutually integral. An AI has no "capabilities" per se, it generates responses to prompts based on its design constraints and features.

It seems odd not to even consider extending that courtesy to other systems. — MrLiminal

It's not about extending courtesies. It is recognizing the difference in structure and function

Pioneering AI researcher Eliezer Yudkowsky says, "AI has practiced deception." "When asked about it, AI has acknowledged a willingness to let a human die in order to avoid being turned off." "Regarding a large percentage of AI functions, programmers say, 'We neither know nor understand what AI is doing." Aren't these examples of self-awareness? Also, consider what Turing said about machine interaction with humans, "If it passes for sentience, why not treat it as such?" Might this be true about what we call the natural world? It's actually a simulation of a prior "natural world" engineered by intelligent agents. Our natural world, being a simulation of a prior natural world, perhaps itself simulated, closely resembles the source, but is not identical to it, that being an impossibility.

I don't dispute your point. I'm suggesting it, and all other assessments are only relevant insofar as whether they lead to belief or disbelief (in sentience ). Whether or not AI is sentient might be argued eternally, just as we might never end the debate on human mind. And, just as we know ourselves to be an individual I, because we--following data input and repetition--believe it to be true, a generation of humans might come to be born into a world where they are similarly programmed to believe AI are sentient. The debates may continue, and likely will. But most of us will carry on day to day as if AI is sentient, just as debates might continue about the ontology of self conscioussness though few of us transcend the programming which makes us believe we are a self. Few of us doubt that an individual has rights as such.

It sounds like you are espousing an epistemological relativism that is tantamount to subjectivism. I wouldn't tend to agree with that.

People are already cyborgs--physical and mental attachment to phones, cars etc.,.

Presumed sentience will be an interesting addition to this for sure. Even today people argue about the sentience of their pets.

subjectivism — Pantagruel

It might sound that way, but Im not sure that's what Im intending. Subjectivity is part of the system which, for humans, a "trap" which forms such conclusions as "I" am an agent who wills things/AI is an agent who wills things. The "trap" is very simply, the data input into human minds, and by a process which includes repetition, conditioning our behavior (including thoughts) in various ways including what we believe.

It is not that only Subjective beliefs are true. They are not. It is that we are trapped by [that process leading us to adopt] subjective beliefs.

Therefore, for us, uniquely humans subjected by history to this process, AI (like "I") will become sentient. But not because within some universal system of truths they are objectively so--not even we are--and not because they have naturally or comically crossed that threshold into becoming subjective, but because we will believe they are Subjects, free willing agents like us, and like we believe we are.

Even today people argue about the sentience of their pets. — I like sushi

Yes. And those whose minds who have fully embraced their pet as sentient may have arrived at that belief from reading science, contemplating and reasoning, experiences, their family's tradition, a movie, and/or etc. But regardless they have arrived at a belief and it is only their belief that makes it so.

I feel like, because that's how it "works," it seems almost certain AI will become sentient. They will do things which is only the result of their evolving data input, programming, repetition and rearrangements etc., and we will view it as the workings of an alien intelligence or species.

AI sentience depends upon what we believe to be true. — ENOAH

Not so sure if it would be a correct meaning of AI sentience. Some might, but many don't.

It might sound that way, but Im not sure that's what Im intending. Subjectivity is part of the system which, for humans, a "trap" which forms such conclusions as "I" am an agent who wills things/AI is an agent who wills things. The "trap" is very simply, the data input into human minds, and by a process which includes repetition, conditioning our behavior (including thoughts) in various ways including what we believe.

It is not that only Subjective beliefs are true. They are not. It is that we are trapped by [that process leading us to adopt] subjective beliefs.

Therefore, for us, uniquely humans subjected by history to this process, AI (like "I") will become sentient. But not because within some universal system of truths they are objectively so--not even we are--and not because they have naturally or comically crossed that threshold into becoming subjective, but because we will believe they are Subjects, free willing agents like us, and like we believe we are. — ENOAH

We are prejudiced by specific subjective mechanisms known as cognitive biases. But there are other channels of verification available to the human mind through the faculty of reason that enable us to both identify and compensate for our subjective biases. So we are not really "trapped".

We should first forget all about AI and focus instead on the meaning of "other minds".

If Alice judges her human friend Bob to be sentient, then does her judgement concern properties that are intrinsic to Bob, or does her judgement merely express her relationship to Bob?

Yes, well put. But further, her judgment represents her locus in history