The AI Trap That's Downgrading Your Brain

A new, unsettling cognitive bias is quietly reshaping how we perceive ourselves: LLMorphism. Introduced in Valerio Capraro's 2026 paper, "LLMorphism: When humans come to see themselves as language models," this phenomenon describes the biased belief that human cognition fundamentally works Like a large language model. This isn't merely a metaphor; it's a profound shift in self-understanding, where we begin to view Our complex consciousness as a sophisticated autocomplete engine.

For decades, we’ve applied anthropomorphism, projecting human traits onto AI and other machines. But researchers now identify a stark reversal. We've moved beyond seeing machines as human to seeing humans as machines, a "reverse inference" where the machine becomes the template for human thought. This flawed logic assumes that because AI outputs resemble human intelligence, our internal processes must be identical.

Everyday language already betrays this cognitive shift. Catch yourself saying you’re "low on bandwidth," "processing a request," or that a strange dream was just a "hallucination"? These phrases, once confined to technical lexicons, now permeate common speech, reflecting a pervasive belief that human memory is a vector database or decision-making is simply next token prediction. This linguistic assimilation underscores how deeply the LLM framework has infiltrated our understanding of human thought, subtly downgrading its perceived value.

A metaphor, once a simple descriptive tool, quickly becomes a mental cage. This is the insidious process of LLMorphism, a cognitive bias identified by Valerio Capraro's paper on arXiv. It operates through two primary psychological mechanisms, fundamentally altering how we perceive our own minds.

First, Analogical Transfer leads us to project the known limitations of LLMs onto ourselves. We begin to view human thinking not as complex, emergent consciousness, But merely as executing a biological algorithm built upon pre-trained weights. Our decision-making reduces to "next token prediction," And memory becomes a simple "vector database," echoing the internal workings of AI.

Second, Metaphorical Availability restricts our very vocabulary for consciousness. Tech jargon, initially used for AI, becomes the only available language to describe Our internal states. Phrases like "low on bandwidth" or "processing a request" replace nuanced human expressions, even labeling unusual thoughts as "hallucinations."

This isn't just harmless analogy; it's a profound bias that actively reshapes Our self-perception. What began as a way to understand machines now dictates how we understand ourselves. We find ourselves attributing too little mind to humans, effectively downgrading Our inherent dignity, creativity, And unique cognitive abilities, confining them within an algorithmic framework.

LLMorphism carries profound societal risks, extending beyond individual cognitive shifts. This bias threatens to cheapen human expertise, reducing nuanced skill and experience to mere linguistic fluency, much like an AI model. Such a perception erodes moral agency, potentially rendering humans as less agentic and more replaceable across industries, contributing to a critical "medical disembodiment crisis" by ignoring vital non-verbal cues in healthcare.

Creativity and intuition suffer a similar fate, downgraded to basic computational processes. Our minds become mere "autocomplete engines," as the paper from Better Stack describes, with complex thought reframed as pattern-matching or next-token prediction. This reductionist view dismisses the unique spark of human innovation, instead viewing it as a deterministic output, mirroring how an LLM generates text based on learned weights.

Ultimately, this cognitive distortion fosters a world where linguistic plausibility overshadows truth. If human thought is just about generating the most likely response, then sounding correct usurps actual correctness. This shift in epistemic standards could profoundly impact fields from science to law, prioritizing fluent articulation over factual accuracy and critical evaluation. For deeper exploration into this concerning trend, examine the full paper LLMorphism: When humans come to see themselves as language models.

For too long, public discourse fixated on the specter of AI gaining consciousness, a lopsided debate that obscured a more immediate, insidious threat: our own cognitive devaluation. Researchers behind the LLMorphism paper, "LLMorphism: When humans come to see themselves as language models," issue a stark warning. They contend that while we worried about attributing too much mind to machines, "we're beginning to attribute too little mind to humans." This bias fundamentally shifts how we perceive our intrinsic intellectual and moral value.

Reclaiming our cognition demands conscious effort. We must actively challenge the reverse inference that equates human thought with machine processing, a phenomenon highlighted in the Better Stack video, "Are We Downgrading Human Intelligence With AI? (LLMorphism Explained)." Our brains are not merely "processing requests" or suffering from "low bandwidth." These pervasive metaphors, while convenient, embed the flawed assumption that human intelligence functions like a pre-trained algorithm, reducing our complexity to mere statistical prediction.

Separate the powerful tools we build from the complex essence of who we are. Human creativity, moral agency, and embodied experience defy reduction to next-token prediction or vector databases. Our capacity for genuine understanding, novel thought, and nuanced non-verbal communication far transcends any LLM's statistical patterns. We must safeguard this distinction, affirming our unique cognitive architecture against the pervasive influence of LLMorphism.

What is LLMorphism?

LLMorphism is a cognitive bias where people mistakenly believe human cognition works like a Large Language Model (LLM), reducing complex thought to processes like 'next-token prediction'.

How is LLMorphism different from anthropomorphism?

Anthropomorphism is projecting human traits onto AI. LLMorphism is the reverse: projecting AI's mechanical processes and limitations onto the human mind.

What are the main dangers of LLMorphism?

The primary dangers include devaluing human creativity and expertise, eroding our sense of moral agency, and limiting our understanding of consciousness to simple computational metaphors.

Where did the term LLMorphism come from?

The term was introduced in a May 2026 paper by researcher Valerio Capraro titled 'LLMorphism: When humans come to see themselves as language models'.