AI produces language without the memory, experience, or stakes of a human mind.

Anti-intelligence describes language generated without a life behind it.

The real shift isn’t smarter machines, it’s language detached from minds.

I’ve previously written that artificial intelligence operates in a different geometry of thought from the human mind. My idea of anti-intelligence is an attempt to describe that difference more precisely.

I've found that this word can easily be misunderstood. When people hear the prefix “anti,” they tend to assume opposition or inferiority, as though the claim suggests that AI is a diminished form of human intelligence. That interpretation misses the intention. The term isn't meant to rank machine cognition below our own; it describes a structural inversion in how language can be produced. Let's be clear. What we're encountering in large language models is not a weaker form of thinking but a fundamentally different architecture operating behind the same medium.

Interestingly, this concept between language and cognition is appearing in the scientific literature as well. A recent paper in Nature Machine Intelligence notes that LLMs often behave in ways that are strikingly realistic in conversation yet remain fundamentally “unhuman” in their underlying structure. The term is well chosen and shares a border with the idea of anti-intelligence. It captures the strange condition where language that resembles human expression emerges from a system that has none of those human experiences.

When Physics Encountered an Inversion

History can offer a practical example. In 1928, the physicist Paul Dirac followed the mathematics of quantum mechanics to a result that seemed rather counterintuitive. His equations suggested the existence of a particle identical to the electron in every measurable respect except one—its electric charge would be opposite. At first, the prediction looked like an odd artifact of mathematics rather than a core feature of nature. Yet four years later, Carl Anderson observed exactly such a particle in a cloud chamber. The positron was real.

Antimatter didn't replace ordinary matter. Instead, it revealed that the structure of the physical world was broader than physicists had previously imagined. Something that initially appeared implausible turned out to be a missing dimension of the same underlying system. I contend that a similar expansion may exist in our understanding of intelligence.

The Architecture Behind the Words

Humans have a cognitive path as thought grows from continuity. It's a vital cluster that, in some way, defines humanity.

Memory accumulates across time

Experience shapes judgment

Decisions carry consequences

The mind isn't simply a processor of language but an autobiographical system in which meaning emerges from lived experience. When we speak or write, our words carry traces of that interior history.

The "anti" part of this is that LLMs operate without that continuity. LLMs generate sentences through statistical relationships within vast collections of information. Simply put, it assembles patterns of language that appear coherent without drawing directly from a lived experience. And this result can sound remarkably thoughtful, but nothing within has ever lived the ideas it expresses.

I sense that describing this as intelligence stretches the traditional meaning of the word. At the same time, dismissing it as a defective form of intelligence fails to capture what makes the technology remarkable. What we are seeing instead isn't another variety of cognition but something structurally different. Anti-intelligence names that inversion. AI uses the same raw material humans do, yet the architecture producing that language runs along a different axis from human thought.

The Intelligence Scale Breaks Down

A lot of the current discussion around AI assumes that humans and machines occupy the same spectrum of intelligence. And then, the debate quickly turns to whether machines will surpass us, or when artificial systems might eventually outthink humans. Those questions feel natural because we imagine a single line where intelligence progresses—and where humans and AI sit along points on that single line.

But that geometry may be wrong. If AI operates along a different axis altogether, the comparison itself becomes misleading. Human cognition brings an experiential cluster: experience, consequence, and judgment into every thought. Artificial systems bring pattern recognition at an extraordinary scale. When these two interact, the results can be productive and even transformative. However, when the distinction between them lessens, AI's statistical coherence can begin to substitute for the slower work of human understanding. That substitution is what I call The Borrowed Mind, a concept I explore more fully in my recent book.

A Category We Did Not Previously Have

Scientific progress often begins with what looks like a mistake. The imaginary number i once appeared to be a mathematical curiosity before it became essential to electrical engineering and even modern imaging technologies. Dirac’s positron initially seemed like an odd prediction before it revealed a deeper symmetry within the physical universe. In each case, the discovery did not invalidate what scientists already understood. It expanded the conceptual space in which that knowledge made sense.

Anti-intelligence may represent a similar expansion. What LLMs reveal isn't that machines have become intelligent in the human sense, but that language itself can now operate within a system that has no mind behind it.

Too human to model: the uncanny valley of large language models in simulating human systems. Nature. 02 March 2026. Y. Zeng et al.

© Psychology Today