I Am An AI Hater

> Words are the most indirect form of perception imaginable. Both Aristotle and Cassirer knew this

What?

Aristotle: There are no contradictions.

Cassirer: “Only when we put away words will be able to reach the initial conditions, only then will we have direct perception. All linguistic denotation is essentially ambiguous–and in this ambiguity, this “paronymia” of words is the source of all myths…this self-deception is rooted in language, which is forever making a game of the human mind, ever ensnaring it in that iridescent play of meanings…even theoretical knowledge becomes phantasmagoria; for even knowledge can never reproduce the true nature of things as they are but must frame their essence in “concepts.” Consequently all schemata which science evolves in order to classify, organize and summarize the phenomena of the real, turns out to be nothing but arbitrary schemes. So knowledge, as well as myth, language, and art, has been reduced to a kind of fiction–a fiction that recommends its usefulness, but must not be measured by any strict standard of truth, if it is not to melt away into nothingness.” Cassirer Language and Myth

I think this is just an extension of the idea that "only 20% of communication is verbal and the rest is nonverbal". We have always understood the limitations of language, most of what is communicated between humans is nonverbal.

we understood the limitations of language, which is why programming was done via...math and logic! Something LLMs seem to absolutely suck at

This is beautiful.

I also had a similar epiphany 3 days ago - once it hits you and you understand it, you can see clearly why LLMs are destined to crash and burn in their present form (good luck to those who will have to answer the questions regarding the money dumped into it).

What will come out of the investment will not justify what has been invested (for anyone who thinks otherwise, PLEASE GO AHEAD AND DO A DCF VALUATION!) and it will have a depressing effect on future AI investment.

> Aristotle: There are no contradictions.

I still don't know what this is supposed to mean, and I am not unfamiliar with Aristotle.

If you don't grasp the basic ideas of paradox questioning the nature of language beginning in presocratics and winding their way through Aristotle, Socrates, Kant Hume and many others and also appearing in Advaita Vedanta (eg Nisargadatta), then I'm afraid staring here isn't really going to help you. Philosophy has ben questioning whether language is valid from the start. https://plato.stanford.edu/entries/aristotle-noncontradictio...

In my experience, it seems like most people believe that they are their thoughts.

This is especially terrible for people with OCD, which seems to be common in this industry. I think it would be a valuable boost to mental health for them to at least explore some of the basic concepts in Vedanta and/or zen.

What amuses me is how much my thoughts seem like a completion LLM while I'm meditating.

Humans understand language at a level no AI does.

We use it to serialize ideas, and we have the ideas independent of language.

AI works on the serialization itself, which is very powerful because of the relationships between ideas are reflected on statistics in the serialization, but it lacks all the understanding, and can't create new ideas with reasonable resources.

You used too many words, and your argument depends on too many people, do you understand all of them?

The way I see it, Aristotle used language as a reasoning tool. Logic inference rules, modus ponens, and so on.

Aristotle was also unaware of the incompleteness problem discovered by Gödel, that no reasoning tool of that type can be complete.

There are fundamental contradictions in the nature of language, it however doesn't make them not useful for the entire experience of daily communication, all of literature, and so on.

Just that there are affirmations that are true, but there is no set of rules that can prove them.

I would point you to Gödel, Escher, Bach for a very nuanced discussion about this topic.

Arguments are all built in neural-syntax specifics. How they externalize as arbitrary points is largely irrelevant, which demonstrates how humans go extinct: confusing the two. The basic 'fact' is words and the conduit metaphor paradox are never resolvable, they are inherently contradictory, which makes words almost entirely irrelevant, and gibberish. AI can never solve this because CS never began at first principles. In essence, AI is the most advanced demonstration of language's total irrelevance.

Now I see your point. You are an absolutist. If something is not 100% perfect and works for all possible cases, then it is immediately worthless. Language being in that category.

I also disagree with your point and your arguments. So many sentences in your response are blatantly false. You can win the Olympics of jumping to conclusions.

Let's start with CS. CS is the set of first principles that are then applied to software. This is because CS is another branch of mathematics, starting with Boolean logic and discrete mathematics.

Language relevance is shown here. We are using it right now. It is not a complete system because some ideas can't be expressed in language and some sentences in a logical system can't be proved or disproved, but the overwhelming majority of sentences are useful.

And everything I have written is based on first principles, you can read about Gödel incompleteness theorem for a start. It applies to LLMs because it applies to all uses of language. Nothing is specific to neural networks.

In fact, go and read about Gödel, because it proves that no logical system is complete, and your worldview seems to be dependent on the outdated assumption that there should be such a complete system. This includes all reasoning systems and all of mathematics.

No my approach is there is no model. Nothing is reducible. It's a neurodynamic approach. The idea of a world model is oxymoronic, the brain doesn't reduce anything to models, making math and logic irrelevant. Nothing you are talking about is really a first principle, how can it be, it's retrofitted using symbols. Yours is a psychodynamic approach, the post-hoc representations brains create is enough for you. You expect reason to be the threshold. I see no reasons for anything, simply actions. The computer I expect uses no math.

btw - this is the Achilles heel to CS "Nothing is specific to neural networks." Pretty fascinating that the illusion of counting, math, algebra will all ne superseded by measurement in analog, a measurement that requires no math, simply differences in syntax. How we code that is up for grabs. How did all these math-heads take control of reality through counting? Really a bonkers group of capitalists had nothing to do except dominate by counted value. Rather insane.

This is an idealized fantasy of language. Language is primarily about patriarchal dominance, control, status, mate-selection, topophilia, and secondarily about communicating ideas. The dark matter of language is expressed in mythological ideas like states, property, law, etc. People are starting to notice we can't solve very simple problems like climate extinction. How the primary forms in language are status oriented.

> Language is primarily about patriarchal dominance, control, status, mate-selection, topophilia, and secondarily about communicating ideas.

I don't agree language is primarily about those things, but I want to point out this is a very human interpretation of language, that no LLM can perform.

As humans don't think in language, ie, there is no direct contact between what we think we think, and how we externalize what we think we think arbitrarily, empirically language is primarily about these biases and only secondarily about communication. This is, again, the Achilles hell of CS, NLP, generative linguistics. It's impossible for anyone to disagree with this function of language since 2016. The role for LLMs to operate language is zilch, as you admit LLMs can't uncover this, train, align this out of function.

“We refute (based on empirical evidence) claims that humans use linguistic representations to think.” Ev Fedorenko Language Lab MIT 2024

The problem of the GEB is it's containment in symbols. Godel was unaware of the the potential for direct perception, and ecological psychology, and coordination dynamics. All three are possible non-contradictory paths to direct perception. Time to put math aside and search for new possibility.

You're being evasive and hiding behind jargon and frankly, nonsensical phrases. As I said, I am not unfamiliar with Aristotle, so don't be shy about making clear and direct claims. If you can't do that, then, I'm sorry, but this some kind is bullshit.

(FWIW, a feature of the Aristotelian logical tradition is that, unlike the modern, Fregean tradition which is indifferent about the relationship between logic and language, it is very much concerned by the logical structures within grammar. From a practical point of view, this makes total sense: we want to be able to evaluate arguments, to clarify arguments, and so on, which are generally given in natural language. Aristotle was also a moderate realist. Language is a reflection of reality.)

The clearest direct claim is if there are no contradictions then language is impossible.

Language is not a "reflection of reality" in any way shape or form: reality is always specific, language is always arbitrary.

We're currently in a neurodynamic/neurobiological overthrow of psychodynamic principles that obviates in presocrats onwards.

The fact is language has nothing really to do with reality and has only to do with subjective biases that arbitrarily perform gibberish in the stead of status-gain, control, etc (pick any primate bias that Aristotle onwards was unconscious to).

“We refute (based on empirical evidence) claims that humans use linguistic representations to think.” Ev Fedorenko Language Lab MIT 2024

btw what this means neurobiologically/neurodynamically is that arbitrary language's only function is to refute itself. That is its only form of reasoning or logic in grammar. It's only a temporary palliative that eventually gets revealed (AI does a great job of this, better that media or the web) that embeds simian bias seamlessly. It demands a direct, concatenated, irreducible form of signaling.