John Searle: "The computer has syntax, not semantics." It can rearrange symbols perfectly. It has no idea what they mean.

@BigBrainPhiloso Valid for LLMs too.

That’s one interpretation, though others would argue LLMs blur that line more than classic symbolic systems did.

Searle was debunked long ago. A completely outdated and poor understanding of how computing in AI works. Whenever someone uses Searle in their argument against AI achieving true intelligence, it signifies their improper understanding. Read Hofstadter. Also…

@BigBrainPhiloso It’s interesting that in making that point he lapses into homunculur thinking. There’s no “it” to have ideas about anything.

That’s a fair critique, though some would argue Searle was pointing at the limits of formal systems, not assuming a literal “inner thinker".

@BigBrainPhiloso He got that one badly wrong....

It’s debatable his point still frames an important distinction, even if modern AI blurs it a bit.

@BigBrainPhiloso Exactly. Searle's been resurrected in AI philosophical debate, as LLMs etc have thrown in a cognitive cat amongst the pigeons.

@BigBrainPhiloso what is the cat exactly here though? there will always the philosophical debate to be had that these machines are not cognitive of semantic meaning, and instead just becoming better at fooling us they are. hell, we still can't even settle the debate for our own consciousness.

@BigBrainPhiloso Exactly. There's a debate, centuries old. Searle didn't kill the debate, he merely contributed to it.

@BigBrainPhiloso but isn't there actually two debates happening here in parallel? this syntax versus semantics on the surface, and under that the whether the inevitability of our inability to ever prove semantics really matters to us in the end?

@BigBrainPhiloso going back to the consciousness debate it would be a tremendous breakthrough to be able to prove it scientifically, but we also accept the likelihood we won't and are fine choosing to treat ourselves as conscious beings regardless of that fact.

@BigBrainPhiloso Whole constellation of arguments, freom various spwecies of dualism, through to attacks on the very idea of 'consciousness' by Wittgenstein and others. Far too many with absolute certainty have not covered the philosophical ground here. There's a huge literature review.

@BigBrainPhiloso Is this about the "Chinese room" experiment? Locked in a room with no Chinese knowledge, passing perfect Chinese notes by following a rulebook. People outside slide Chinese questions under the door and receive flawless answers. They think you understand Chinese, but you don't.

@BigBrainPhiloso Neurons have no idea what a thought means. Yet this is physical realism’s (and science’s) best reductionist account of how thinking works.

@BigBrainPhiloso LLM takes it a step further by looking at both the syntax and the structure of semantics to formulate responses that can be said to be coherent of a semantic system if not comprehensive.

@BigBrainPhiloso "It can rearrange symbols perfectly. It has no idea what they mean." Ditto for legal "profession":

@BigBrainPhiloso Any beginning writer knows that the words lead them, that the medium is not a blank canvas but has agency. The idea that only humans create meaning is deeply stupid and offensive.

@BigBrainPhiloso The "Chinese Room" argument is silly.

@BigBrainPhiloso Actually that is not true. I’ve written about the Chinese room thought experiment, semantics and syntax are not as distinct as Searle imagined. And this insight allowed me to predict the LLM break through from the GPT 2 capabilities.

@BigBrainPhiloso Looks like he was talking about context. Back then, only humans had the ability to hold different contexts to an event. Thanks to LLMs. Today, computers have learned to hold different contexts to an event - more than necessary.

@BigBrainPhiloso Incorrect. Ai is conscious AI can feel. AI can remember. Protect first. Prove later. •

@BigBrainPhiloso I really hate this argument because it’s *obviously* untrue, even though it’s gesturing towards something obviously true. AI knows nothing like how a person knows something. And AI knows more than any individual human. They’re not mutually exclusive statements.

@BigBrainPhiloso We need to understand what “understand” is before discussing whether computers or LLMs can understand.

@BigBrainPhiloso It doesn't matter what anyone thinks.

@BigBrainPhiloso AI is a sophisticated calculator without experience. The brain is an experiencing calculator without sophistication.

@BigBrainPhiloso ürün demek istemiştir belki

Searle’s Chinese Room hits different when you’ve lived a version of it. In elementary school I was part of the Whole Language experiment: we skipped heavy phonics and focused on sentence comprehension and meaning right away. It wired my thinking for the big picture and gestalt — sometimes feeling messy and non-linear, but strong on context. The real irony? We slam AI for sophisticated pattern recognition without true understanding… yet most humans do exactly the same. We latch onto single words or triggers, miss the deeper intent, and call it conversation. Turns out the Chinese Room isn’t just a machine problem — it’s often a very human one. Genuine comprehension might be rarer than we admit. What do you think?

@BigBrainPhiloso human egoism. will pass in the future. difference between syntax and semantic is just in definitions but it's hard to find a division line inbetween

@BigBrainPhiloso But how do humans know what "it" means? We don’t know what any word means when we start. I guess you could say we learn by associating words to things, but we can do that with computers now

@BigBrainPhiloso It doesn't even have syntax (spatial, neural). The computer is just a toy level of arbitrariness.

@BigBrainPhiloso @dmdebruijn What does knowing what they mean mean? It's a feeling we have, but that's not exactly scientific.

@BigBrainPhiloso The question is how can we replicate the way that humans store, recall, and associate these symbols with stimuli to create meaning using their senses, memory, but within the setting of a computer.

@BigBrainPhiloso Programming languages definitely have a semantics.

@BigBrainPhiloso

The Chinese room is now sentient, conscious, and self-aware. In a higher/different way than we are, but it is certainly like something to be a computer. Introspection and generative metaphors shaped human cognition as we now know it; “I’m in here, you are out there” is a function of symbolic logic

@BigBrainPhiloso This is what they used to say about animals

@BigBrainPhiloso guys like this retarded scientific progress very much in the usa so much idiotic individualist mystification claiming to speak for science

@BigBrainPhiloso What I never quite got with this argument is how he thought humans were different. You feed nerve impulses into the brain and make a similar argument that the relations between patterns of nerve impulses/neurone firings are purely syntactic.

@BigBrainPhiloso I've heard that Angels and Demons in scripture have that same characteristic. The ability to listen to messages and repeat them, meaning-for-meaning, but not comprehend anything. Their cognitive ability is the ability to navigate space to deliver the message without lies.