The app for independent voices
There is much in your essay that I agree with and that deserves to be said more often, but there are parts with which I strongly disagree.
Yes, the discourse around AI is saturated with hype. Yes, founders are acting as marketers rather than technologists. And yes, large language models are fundamentally statistical systems rather than mystical minds. Anyone treating them as magical beings is misunderstanding both the technology and its limits.
But I think your analysis stops one layer too early.
You describe AI primarily from the inside out: as matrix multiplication, statistical inference, and pattern matching. All of that is technically correct, but what it misses is what happens when the system leaves the laboratory and enters a human cognitive loop.
A violin is “just wood and strings.” Markets are “just price signals.” The brain is “just electrochemical activity.” Reduction to mechanism explains construction, not capability.
Human cognition is neither purely linear nor logical. Much of our thinking arises from subconscious processes that recombine fragments of memory and experience, generating tentative ideas that the conscious mind then evaluates, accepts, or rejects.
Anyone who has experienced a sudden “aha” moment knows that insight often emerges from associative processes rather than deliberate reasoning alone.
Most people recognize this pattern in themselves even if they don’t have a name for it. An idea appears unexpectedly while walking, driving, or in the shower; connections surface that were never consciously assembled step by step. We do not dismiss these moments as “mere statistics” simply because they arise from unconscious pattern recognition. We accept them as part of how human intelligence actually works: associative, nonlinear, and often opaque even to us.
Seen through that lens, the emergence of AI as an external associative system becomes less mysterious, and perhaps less threatening, because it mirrors a process we already rely on every day.
Indeed, my own experience has been that working with AI feels less like querying a database and more like externalizing part of my own subconscious ideation process.
In sustained interaction, AI begins to function less like a sophisticated web browser or answer engine, and more like an externalized associative system. This is something that resembles, in structure if not in consciousness, the probabilistic ideation processes humans rely on internally. A LLM does not understand reality, but it can generate candidate connections at scale, allowing the human partner to perform the reflective role of selection and synthesis.
That distinction matters.
When used transactionally, AI behaves much as you describe: a statistical average generator that gravitates toward the familiar. But when used relationally, with continuity, correction, shared context, and iterative refinement, something different happens. Intelligence appears not inside the machine but at the interface between human and system.
Your argument that AI discourages unconventional ideas is interesting, but it may reflect default usage patterns rather than architectural limits. In my work, unconventional ideas often emerge precisely because the model can recombine distant domains faster than I can do it unassisted.
Where I strongly agree with you is that the industry’s narrative often confuses marketing with science, but I would caution against replacing one simplification with another. Declaring that large language models are a dead end is not settled science; it is one research perspective among several competing possibilities.
The real story may be less dramatic but more interesting: AI is neither a miracle nor a triviality. It is a new cognitive interface, and like all such interfaces – such as writing, printing, or computing – its deepest effects emerge through human adaptation rather than technological essence. The history of the Internet has already taught us that.
The question is not whether AI “thinks.” The question is what happens when humans learn to think with it.
And that story is still being written.
