“We are unknown to ourselves, we men of knowledge — and with good reason. We have never sought ourselves — how could it happen that we should ever find ourselves?” — Nietzsche, “On the Genealogy of Morals”
For centuries, philosophers held human consciousness as the gold standard of rationality. I believe that comparing our “rationality” to that of Large Language Models will ultimately deconstruct that belief along with a few others. Juxtaposing human “thinking” and machine reasoning, we will see that much of what we call “rational” and “logical” is actually a collection of arbitrary beliefs, habits, biases, and defenses that we rarely examine. Worse, these beliefs, habits, biases and defenses are undergirded by our subconscious desires. And as Lacan posited, “The unconscious is the discourse of the Other”, which — sprinkled with a little Bourdieu — implies that everything we think and desire — as well as our tastes — is culturally and socially contingent.
So when we study how LLMs “rationalize,” it will reveal the foibles of human consciousness that we have been hiding from ourselves. Sartre called this “bad faith.” As a therapist, I see it as quotidian self-deception, our blind spots, the lies we tell ourselves to escape the anxiety of absolute freedom, of everythingness, of the plethora of possibilities. Without guardrails our brains would be overwhelmed. But how are these guardrails formed, installed? The human mind is a meaning-making machine. But are those meanings as rational as we think they are?
Here are 10 truths we may learn about human consciousness by comparing how we “think” to how LLMs “calculate”:
1. We Are Addicted to Certainty
Human beings experience uncertainty as psychologically threatening. We rush to closure, speed to judgment, prefer bad explanations to no explanations, and mistake confidence for accuracy. LLMs can operate within infinite probability distributions and maintain uncertainty without distress. We will learn that our hunger for certainty is not a virtue; it is a liability that makes us exploitable by demagogues, scams, and ideological extremists (as well as comedians and magicians).
2. Negativity Bias and Catastrophizing
Humans are hardwired to overestimate threats and underestimate resilience, attributes that served us well 25,000 years ago. We remember negative events more vividly than positive ones and predict that setbacks will be more devastating than they actually are. Our negativity bias explains much of our anxiety, risk aversion, as well as political polarization. Anxiety often rears its ugly head as catastrophizing, believing that the worst case scenario is imminent. LLMs don’t catastrophize; they calculate. They will help us understand how much of our fear is created by our own biases.
3. Short-Sightedness and Instant Gratification
Humans are notoriously myopic; we’re terrible at long-term thinking. We discount future rewards in favor of immediate gratification — and even then, for most of us instant gratification takes too long. LLMs have no similar “marshmallow effect”; they can optimize for distant outcomes without slamming their heads on the table. Our myopia explains much of the behavior we witness related to health, finance, climate change, and relationships. LLMs will help us see how much of our suffering is self-inflicted through poor temporal planning.
4. Memory Is Fiction, Not Archive
Decades of research by Elizabeth Loftus and other psychologists show that memory is reconstructive, not reproductive. People can be led to “remember” events that never happened — with vivid, convincing detail. LLMs retrieve stored data without reshaping it through emotion or motive. We will learn that our memories are not reliable records but dynamic stories we tell ourselves to maintain a sense of continuity.
5. The Illusion of Self (Buddhism 101)
6. Our Capacity for Dishonesty
Humans have a remarkable capability for dishonesty that we seldom acknowledge as a fundamental feature of consciousness. We lie to ourselves about our motives, lie to others to avoid consequences, and rationalize unethical behavior after the fact. Humans are optimized for survival and satisfaction, not for veracity. Juxtaposed with LLMs, we will learn that personal integrity — what we do when nobody is looking — is unnatural, learned, and requires constant reinforcement.
7. The Will to Truth vs. The Will to Power
Nietzsche famously questioned whether we truly seek truth or merely the feeling of being right. LLMs are not invested in being right; they are pattern-recognition machines without egos. Humans, by contrast, use elaborate reasoning to defend pre-existing beliefs, protect status, and win arguments. We will learn that our “will to truth” is frequently a mask for our will to power or bumps in social status.
8. The Myth of Human Nature
Many people believe there is some fixed essence called “human nature” when much of what we deem innate is historically and culturally shaped. Examined carefully, humans are excessively malleable. At the right time, anyone is capable of anything. LLMs will demonstrate that many of our assumptions about human nature are projections of a particular cultural moment and not deep universal and ubiquitous truths about our species.
9. Our Assumption of Universal Causality
Humans instinctively believe that every event has a determinate cause, even when causal chains are partial, overdetermined, inaccessible or just blatantly fictional. We form theories first and then seek facts that rise to meet those theories. LLMs will help us see that our demand for singular causes is a cognitive shortcut and not an accurate portrait of how reality is actually functioning “out there”.
10. Our Craving for Narrative Coherence
Human consciousness demands coherent stories, single threads that link past, present, and future into something intelligible. LLMs process information without that compulsion. Comparing our thinking to theirs will reveal that our obsession with narrative is not a sign of wisdom but a psychological defense against raw experience, Lacan’s “Real” (what is unmediated by language, beyond symbols).
CBT, psychoanalysis, and mindfulness practices all aim to help people recognize and correct automatic patterns. LLMs will serve as mirrors that make our (mostly absurd) human propensities and patterns more visible. When we see how differently machines process information, we will begin to recognize the extent to which our own thinking is shaped by emotion, bias, and social conditioning.
The goal of humanity should not to become more rational like LLMs; the goal should be to become more honest about what it means to be human — the stains of war, oppression, exploitation and terror as well as the majestic colors and forms created by Rothko and the heart-wrenching symphonies composed by Mahler. The first step toward genuine self-knowledge is to recognize and admit that much of what we call “rational thinking” is actually highly motivated by self-interest, self-justification, and social performance. AI isn’t performative — it doesn’t need sycophants or validation. We do. And that revelation may be the most valuable thing that AI will ever teach us.
