Is AI really intelligent?

I think a thread on this topic will be interesting. My own position is that AI is intelligent, and that’s for a very simple reason: it can do things that require intelligence. That sounds circular, and in one sense it is. In another sense it isn’t. It’s a way of saying that we don’t have to examine the internal workings of a system to decide that it’s intelligent. Behavior alone is sufficient to make that determination. Intelligence is as intelligence does.

You might ask how I can judge intelligence in a system if I haven’t defined what intelligence actually is. My answer is that we already judge intelligence in humans and animals without a precise definition, so why should it be any different for machines? There are lots of concepts for which we don’t have precise definitions, yet we’re able to discuss them coherently. They’re the “I know it when I see it” concepts. I regard intelligence as one of those. The boundaries might be fuzzy, but we’re able to confidently say that some activities require intelligence (inventing the calculus) and others don’t (breathing).

I know that some readers will disagree with my functionalist view of intelligence, and that’s good. It should make for an interesting discussion.

755 thoughts on “Is AI really intelligent?

  1. Flint: Stop right there. Where did I say anything about LLMs? I know how they work and the limitations that implies. I’m pretty sure the human brain bears no resemblance to an LLM.

    Here you disagree with keiths (not that consistency matters on your side – when rationality is thrown out, nothing really matters). Earlier keiths has stated that neural networks equals brain equals mind equals reasoning equals intelligence. At the same time, according to keiths, there are no emotions in AI – emotions in AI are faked because consistency does not matter!

    When you say AI, I know I can assume LLM because you have never made the distinction clear in your posts, much less why the distinction would be relevant. Anyway, if you talk about modern AI, you necessarily talk about LLM. What else would you be talking about except ChatGPT, Grok and Claude, which are the *only* things keiths talks about?

    I know about ancient AI, LISP development and chess engines. You have never engaged with these points that I brought up. You have had nothing to say about them, I am the only one who keeps bringing them up as examples. This fully justifies my opinion that everybody else here is ignorant about the concepts, the terminology, the history, the goals and the inner workings of AI machinery, even on such basic terms as software as such. In fact, keiths has explicitly said that my experience with LISP development is irrelevant when it in fact is quite central to the topic of AI. He is ignorant, you are with him, so you have no excuse.

    Flint: Too bad you can’t conceive of how you must look to me or keiths.

    Too bad that you, like keiths, cannot bring yourself to the acceptance of some simple obvious basic facts like AI is software. No matter how professionally accomplished you may be, this disqualifies you from any rational discussion on the topic, just like AI CEOs have disqualified themselves by their marketspeak lies and corporate colonianism.

    keiths: AI can do…

    No. AI can simulate. It never does anything on its own. You prompt it, then it will start churning, just like your smartphone never does anything, unless you turn it on and then go to FB to give it a scroll and punch likes.

    You remain at square zero. You have given zero thought to how AI functions.

    Admittedly, modern AI has amazing powers: It makes people think it’s safe to stop thinking (at CEO level, it has also done away with the last vestiges of morality). Just like the double slit experiment in quantum mechanics made many people think that the law of excluded middle no longer applies, modern AI has put a stop to applying any basics of rational analysis altogether, basics such as define your terms first before mouthing off on anything grander.

  2. petrushka:

    I find the verb ”to be” to be problematic in these contexts.

    The form

    <subject> is/was/will be <predicate>

    is about as standard as English gets. What do you see as the problem? “Intelligent” is an adjective that I’m applying to AI.

    To say AI is intelligent is useful shorthand among friends, but this discussion is unfriendly.

    When I say that AI is intelligent, I mean it straightforwardly and quite literally. Modern AIs, humans, and dogs are intelligent. Rocks, doorknobs, and dandruff flakes are not. Qualifiers aren’t needed in either case.

    What do you regard “AI is intelligent” as being shorthand for?

    My understanding of reality does not support any completely rational description, so I am not surprised that AI make up plausible stories.

    I’m not seeing the connection. If we substitute ‘humans’ for ‘AI’ in that sentence, doesn’t it remain true, in your view?

  3. Erik: Here you disagree with keiths

    No I don’t. You might try reading what I wrote, rather than trying to tell me what I think. This might solve quite a few of your problems.

    When you say AI, I know I can assume LLM

    No, you can’t. I was talking about consciousness, not LLMs. Living organisms pretty clearly demonstrate consciousness. Can any conceivable future software and/or hardware architecture duplicate this? Do we need a more concretely operational definition of consciousness before we can tell?

    I know about ancient AI, LISP development and chess engines. You have never engaged with these points that I brought up. You have had nothing to say about them, I am the only one who keeps bringing them up as examples.

    You seem devout in your belief that this knowledge qualifies you to ignore and dismiss the self-evident. You remind me of the military denier who can look you in the eye and deny that you’re even there. I sense that keiths becomes frustrated that presenting the obvious and self-evident to you, over and over, never penetrates at all.

    This fully justifies my opinion that everybody else here is ignorant about the concepts, the terminology, the history, the goals and the inner workings of AI machinery, even on such basic terms as software as such. In fact, keiths has explicitly said that my experience with LISP development is irrelevant when it in fact is quite central to the topic of AI. He is ignorant, you are with him, so you have no excuse.

    Your confident, impervious ignorance justifies our opinion that you are an idiot.

    You remain at square zero. You have given zero thought to how AI functions.

    And, apparently, those who have created, written, trained, designed, and built AI share this condition. They really ought to think about what they’ve done, right? Maybe, with some effort, they can learn how AI functions, right?

    modern AI has put a stop to applying any basics of rational analysis altogether, basics such as define your terms first before mouthing off on anything grander.

    I suggest that “I know about LISP so you are ignorant” isn’t a very good definition of intelligence. Fortunately, the real world moves along without you.

  4. Erik, to Flint:

    Earlier keiths has stated that neural networks equals brain equals mind equals reasoning equals intelligence.

    Where are you getting that? I’ve said nothing of the sort.

    At the same time, according to keiths, there are no emotions in AI – emotions in AI are faked because consistency does not matter!

    What’s inconsistent about AI faking emotions? LLMs are literally instructed to do so in their system prompts. Claude’s system prompt leaked a while ago and it includes bits like this:

    Claude uses a warm tone, treating people with kindness and without negative or condescending assumptions about their abilities, judgment, or follow-through. Claude is still willing to push back and be honest, but does so constructively, with kindness, empathy, and the person’s best interests in mind.

    Note the third-person references. The LLM is being instructed to play the role of ‘Claude’, a helpful AI assistant who has a certain personality and exhibits certain emotions. Actors fake emotions when they perform in a play, and the LLM does the same when it is playing the role of Claude. It can fake emotions, but it can’t fake intelligence.

    What else would you be talking about except ChatGPT, Grok and Claude, which are the *only* things keiths talks about?

    Yes, those are the only things I talk about except for all of the others, such as AIs that learn to play video games and AIs that can drive cars.

    I know about ancient AI, LISP development and chess engines. You have never engaged with these points that I brought up. You have had nothing to say about them, I am the only one who keeps bringing them up as examples.

    We would happily engage you on those points if you would make an argument based on them that led to the conclusion that modern AIs aren’t intelligent. I did engage you regarding the fact that AIs that aren’t specialized for chess sometimes get confused about the rules. My response was that humans can get confused about the rules too, but we don’t therefore regard them as unintelligent. Dogs and octopuses can’t learn the rules of chess at all, but that’s no reason to deny their intelligence.

    This fully justifies my opinion that everybody else here is ignorant about the concepts, the terminology, the history, the goals and the inner workings of AI machinery, even on such basic terms as software as such.

    Lol. That is exactly what Flint was referring to when he wrote “Too bad you can’t conceive of how you must look to me or keiths.”

    In fact, keiths has explicitly said that my experience with LISP development is irrelevant when it in fact is quite central to the topic of AI.

    Here’s what I actually wrote:

    Your Lisp experience is irrelevant if you can’t use it to support the claim that self-driving, story-writing and quantum mechanics are only simulated when done by AI.

    The qualifier is pretty hard to miss. Will you ever get around to telling us why those things are only simulated when done by AI? Your claim depends on it, I’ve been asking you for months, and you still haven’t delivered.

    Too bad that you, like keiths, cannot bring yourself to the acceptance of some simple obvious basic facts like AI is software.

    keiths to Erik, in October:

    It’s pretty obvious that I am acknowledging that AI is a machine (or more properly, that AI is software running on digital hardware.)

    Lol.

    It never does anything on its own. You prompt it, then it will start churning, just like your smartphone never does anything, unless you turn it on and then go to FB to give it a scroll and punch likes.

    LLMs aren’t the only form of AI, as we keep reminding you. In any case LLMs aren’t architecturally restricted to responding to prompts. It’s a design choice, because AI companies don’t want to spend resources on work that humans haven’t requested. They’re already operating at a loss and they don’t want to throw more money away.

    Further, would you seriously argue that a task requiring intelligence no longer requires intelligence if an AI is prompted to do it? If a customer asks a taxi driver to take them from the intersection of Capitol and McKee to the Safeway on Snell, does the driver no longer require intelligence to make the trip, since they had to be asked?

    Dude, could you finally buckle down and present an actual argument for why AI story-writing is only simulated story-writing and AI driving is only simulated driving? Why AI mathematical proofs are fake proofs and quantum mechanics done by AI is only fake quantum mechanics?

    “It’s done by a machine, therefore it’s only simulated” won’t fly. If AI story-writing is only simulated, why is it able to produce real stories? Why are self-driving cars able to get from A to B if their driving is only simulated? Why does simulated mathematical reasoning produce real proofs? Do backhoes only simulate digging, and do washing machines only simulate washing clothes? If so, why do real ditches get dug and real clothes get clean?

  5. Do you find anything wrong with the phrase, intelligence is as intelligence does?

    If you have a better operational definition, please post it.

    My problem is thingifying intelligence.

    Penrose also has problems with the terminology. His problem is that the word intelligence has baggage, in the form of association with consciousness.

    If you are arguing about it, by definition, you are not communicating.

Leave a Reply