I hope you don't mind if I sit this one out.

My argument to support the provocative title of this discussion is: AI is indeed intelligent in that it is able to find patterns in huge amounts of data but there is no way AI could reach to judgements like we humans can.

AI operates under a totally different set of motivations from real intelligence, and thereby is intelligence separated from particularity. It's like an anthropomorphic idea of human intelligence, hence the 'artificial' part. At a certain point, we must decide: is the expediency we are seeking also the idea of expediency? Which is more expedient, the factory that produces artwork as artificial production or the artist themself?

AI is indeed intelligent in that it is able to find patterns in huge amounts of data but there is no way AI could reach to judgements like we humans can — Pez

I can make up a question with a scenario in which a rule must be applied that most AIs would be able to answer correctly. But I don't think that is what is meant by judgement here. AIs can repeat something based on the huge amounts of data their alrogithm has been fed, but they can't think — is what you are saying. Some people do believe that some AIs think.

But to leave decisions regarding therapy entirely to AI or replace judges and jurors by AI would be fatal. — Pez

Leaving it to humans is fatal too.

If we define “stupidity” not as lack of intelligence but lack of judgement, one might justly say that Artificial Intelligence is intelligent but still stupid in its own way. I am referring here to the Kantian concept of “Urteilskraft”. According to this philosopher intellect (Verstand) is the power to create rules whereas judgement is quite a different faculty, enabling us to decide, whether a specific rule applies to a single instance.

Rules are always common to many different cases. To find out, if a singular case applies to a rule, judgement is necessary. There is no way to find this out by exerting another rule, as this results in an infinite regress. So my argument to support the provocative title of this discussion is: AI is indeed intelligent in that it is able to find patterns in huge amounts of data but there is no way AI could reach to judgements like we humans can.

Such discussion might sound academic, but it has impacts on the proper use and possible dangers associated with this technology. AI is still in the state of infancy right now, but it is nevertheless amazing what it is able to do. No doubt, it can be of invaluable help in finding medical diagnosis or related precedences in jurisdiction. But to leave decisions regarding therapy entirely to AI or replace judges and jurors by AI would be fatal.

The road to our house is quite steep, winding and narrow, but once in a while big trucks get stuck there because the drivers relied entirely on their navigation system instead of applying common sense as well. To put our faith solely on AI might get us into the same situation unless we properly exert judgement and check, if the outcome is plausible at all. — Pez

I think that you're on to something. From my limited understanding of Kant, and as you pointed out, the intellect or understanding provides concepts and when we judge something we are determining whether or not that particular falls under the concept. The concept itself, though, provides the rule for determining whether or not the particular is captured by the concept. I believe that ai can more or less do this now, as Lionino pointed out,... At least in theory, with the right systems and data.

But human understanding can also supply its own concepts in instances where one isn't supplied beforehand. And human reasoning can form judgements about judgements made by the intellect. If an ai could ever form its own concepts we could perhaps say that it understands those concepts. If it could form judgements based on its understanding of those concepts, then perhaps we could say the ai can reason. ai is clearly nowhere near this, but its foundation is there in principle.

Your word's but a whisper, anyway.

Thank you. Was hoping I wouldn't be left hanging there.

Sperm in the gutter, so to speak.

I wonder who we left behind, and who else is cultured.

One would assume Pez, at least, is cultured. The others? Well, we can't make them think.

I supose I'd best add a bit of content.

The recent New Scientist has an article about AI empathy.

...human participants and ChatGPT were given descriptions of different scenarios and asked to write short, compassionate answers. When other participants rated the various responses, they scored the AI responses as highest for empathy.

I find that somewhat surprising.

And this:

Concerns over the danger of machines that can “read” us but don’t care about us are more than theoretical. In March 2023, a Belgian man reportedly died by suicide after six weeks of discussions with an AI chatbot. Media outlets reported that he had been sharing his fears about the climate crisis. The chatbot seemed to feed his worries and to express its own emotions – including encouraging him to kill himself so that they would “live together in paradise”. Pretending at empathy to too great a degree without the common-sense guard rails that a human is likely to offer can, it appears, be lethal.

The AI's don't know how it feels...

It is not only about judgement. AI couldn't reach higher stated of mind. Or psychedelic state of mind. It can never be achieved. Humans regardless it is scientifically proven or not , experience a higher state of mind or atleast a state of bliss. It won't be acquired by AI. I think humans could focus on advancing human beings rather than advancing AI it is a big mistake. By making AI and robots we are limiting ourself. Making us weak and makes it hard for us to evolve into better beings. AI will never experience an existential crisis those things are only experienced by humans .

AI is indeed intelligent in that it is able to find patterns in huge amounts of data but there is no way AI could reach to judgements like we humans can. — Pez

Agree. I’ve engaged in dialogues with ChatGPT almost daily since its release and am frequently amazed and impressed with the quality of the output. But I agree with your basic point. The reason is, human judgement ultimately relies on what really matters, what we think is important or salient, among other things. And the point about the distinction between rule-following and judgement is also true.

Actually on the point of becoming too reliant on AI, I’ve noticed that since GPS systems have become common, I have over-relied on them on many occasions, even allowing them to direct me via a route I kind of instinctively felt was the wrong one more than once. And I’ve noticed I will now routinely turn on GPS instead of relying on my own navigation skills like I used to. I can really see this happening in an AI-dominated future. ‘But Siri told me…..’

Leaving it to humans is fatal too. — Lionino

Sadly enough You are right and many a dictator even nowadays proves it. That is why, in democratic systems at least, not a single person alone can make decisions concerning a whole population. AI cannot be made responsible for its decisions, but a scoundrel very well.

:grin:

but a scoundrel very well — Pez

Is he? Some people genuinely can't help themselves. That should not stop us from jailing them, but why blame the fish for swimming and the bear for brutally mauling the off-trail hikers?

AI couldn't reach higher stated of mind. Or psychedelic state of mind — Abhiram

You can't make AI have psychedelic experiences because it does not have chemicals in itself. But the argument of whether computers can have something alike psychedelic experiences is the whole argument about whether AI can think (have reason) or not.

It is a bit cheap to proclaim that AI will never be able to achieve something or other because it hasn't done so yet, or just because. A more substantive argument is needed for such theses to be taken seriously.

The history of game-playing AI is a case in point. It wasn't so long ago that chess experts confidently predicted that AI would never be able to beat a grandmaster. At best, they argued, extrapolating from chess computers of the day, they would hold their own against most players through sheer brute force calculation and programmed strategies, churning out dull games. Others maintained that even if AI chess mastery were achievable in principle, mastering Go would just be mathematically impossible. Well, we know how that turned out. And the point of the story is not to extrapolate further AI successes. I am not even an AI optimist - I am agnostic about its future. The point is that arguing for the impossibility of a future development in a field (especially lacking any relevant expertise) is an ungrateful endeavor. (If Agnostic Meta-Induction hasn't been coined yet, I am coining it now :smile:)

I get the impression that many of those who insist that AI will fail are rather anxious for it to fail, so this is may also be a kind of wishful thinking.

Rules are always common to many different cases. To find out, if a singular case applies to a rule, judgement is necessary. There is no way to find this out by exerting another rule, as this results in an infinite regress. So my argument to support the provocative title of this discussion is: AI is indeed intelligent in that it is able to find patterns in huge amounts of data but there is no way AI could reach to judgements like we humans can. — Pez

It is odd that you talk of rules as being contrary to common-sense judgement, because rule-based AI is currently out of fashion. So-called expert systems of the early days of AI were rule-based. They showed promise, but proved to be limited and have since been largely abandoned. But some prominent AI researchers now argue that further advances in AI will require an infusion of rule-based reasoning. And part of their argument is that common knowledge and common sense that AI often lacks is very much rule-driven (rule-of-thumb driven, one might say).

Actually I was trying to explain that AI couldn't experience higher state of mind. Lot of people cannot understand that and some doesn't believe in that. Somewhat closer effect is a psychedelic experience that's why I used that.

It is a bit cheap to proclaim that AI will never be able to achieve something or other because it hasn't done so yet — SophistiCat

Thank You for Your comment, it might lead to some clarification about what is at stake here.

I have to admit my statement (“ there is no way AI could reach to judgements like we humans can”) is a bit bold and so is the title of this discussion. You are surely right that there have been predictions in history being utterly ridiculous today, for instance the claim of some scientists that no device heavier than air could ever be able to fly.

So let us assume humans in a far future are able to construct robots so perfect, that their behaviour is in no way distinguishable from the behaviour of humans. This could lead to questions like “are these machines conscious?” or “do they have an inner life like we have?”. Nevertheless we could still have our discussion about “judgement” at large. It is a matter of “principle” so to speak and principles depend on our world-view.

There have been philosophers stating, that “in principle” we cannot know anything about things independent of our faculties of knowledge at all. Even if You do in no way consent to such idealism, You cannot prove that it is wrong. All You can do is construct Your own system of ideas and provide convincing arguments for it in contrast to the other opinion.

The exciting feature of our forum is that we can discuss conflicting arguments and maybe learn from diverging points of view.

Is he? Some people genuinely can't help themselves — Lionino

This would lead us to a different type of discussion regarding determinism, causality and personal responsibility (ability to have a choice). Maybe I can provide something like that in the future.