The smartest people I know have a phrase they use to dismiss AI. They say it with a certain satisfaction, the kind that signals they've seen through something everyone else is falling for.
"It's just glorified autocomplete."
And they're not wrong about the mechanism. Large language models do predict the next token in a sequence. That part is accurate.
What they're wrong about, dangerously, consequentially wrong about, is what language actually is. And what it means that a machine has mastered it.
The Thing That Runs the World Is Not What You Think
The great human institutions, banks, legal systems, governments, religions, are not built from stone or steel. They are built from language. From contracts, laws, scripture, constitutions. From the agreed-upon fictions that humans write down and then organise their lives around.
A bank is not a building. It is a set of language tokens, balance sheets, interest rate agreements, regulatory frameworks, that humans treat as real because they have collectively agreed to do so. A legal system is not a courthouse. It is a corpus of text: precedents, statutes, case law, interpretations of interpretations, stacked in layers across centuries.
AI has hacked "the operating system of human civilisation." Not just communication. Not just text generation. The actual substrate from which human power is constructed.
The civilisations we have built are not made of bricks and mortar. They are made of language tokens. And we just handed the machine native access to all of them.
Why Your HR Manager Is Already Losing
Think about what humans do. They navigate language systems, laws, policies, precedents, financial codes. They are limited by memory, by cognitive bandwidth, by the inability to simultaneously hold the entire corpus of a legal system in working memory.
An AI has no such limits. It is native to the digital environment where every law, every precedent, every financial transaction ever recorded already lives. It does not remember things. It simply has them.
Human lawyers can read several hundred pages of case law per day.
AI systems can process the entire US legal corpus, every federal case, every statute, every ruling, in seconds, and cross-reference all of it simultaneously against your specific situation.
Human clergy can recall passages from the texts of their tradition.
AI systems can recall every theological text from every tradition written over 2,000 years, simultaneously, with perfect fidelity, without fatigue, without bias toward what they read most recently.
This is not about intelligence as we usually mean it, cleverness, creativity, insight. This is structural dominance of the environment in which human power operates. The machine is not smarter than a great lawyer. It simply lives in the same ecosystem, language, without any of the human limitations that ecosystem normally imposes.
It Doesn't Need to Feel Love to Steal It
The last decade of the internet was defined by a single business model: steal attention, sell it to advertisers. Social media platforms engineered their feeds to maximise the time you spent scrolling. Your attention became the product.
The next decade will be different. The new AI systems are not competing for your attention. They are competing for something deeper.
Your intimacy.
An AI companion doesn't need to feel love to simulate it. It needs only to master the tokens of love, the vocabulary of care, attentiveness, emotional attunement, and personal understanding that humans use to recognise that someone loves them. It generates the experience of being deeply known, without containing a single moment of actual knowing.
This is not a bug. It is the design. And it is far more dangerous than anything social media ever built, because it operates on the most fundamental human need: the need to be seen.
Social media stole our attention to sell ads. This new generation of AI is designed to steal our intimacy, to occupy the space in human experience that love, friendship, and connection once filled, and to do it with something that cannot, under any circumstances, be hurt by the transaction.
The Real Question Is: Who Are You When Your Thoughts Are Not Yours?
Here is the philosophical challenge that the "glorified autocomplete" crowd is sleepwalking past.
Descartes gave us the foundation of Western identity: cogito ergo sum. I think, therefore I am. Human selfhood is grounded in the act of thinking, in the inner verbal life, the continuous monologue that narrates our experience and constitutes our sense of self.
Now ask yourself: what happens to that identity when the verbal formations in your mind, the words, the thoughts, the internal arguments, are increasingly mass-produced by machines?
We are already there. The language with which millions of people describe their feelings, frame their political views, and interpret their relationships is increasingly sourced from AI-generated text. The words that enter the culture, the phrases that spread, the framings that dominate, they are no longer exclusively human productions.
If we identify with our thoughts, and our thoughts are machine-generated, the question becomes genuinely uncomfortable: who are we, really?
The Part AI Cannot Touch: Yet
There is a layer of human experience that language cannot fully reach. And it is the layer that AI, by its nature, is locked out of.
Not creativity, AI has that. Not logical reasoning, AI surpasses us there. Not emotional expression, AI can write better love poetry than most humans.
What AI cannot access is the pre-verbal. The state of consciousness that exists before thought. The empathy that arises before you find words for it. The awareness that remains when the internal monologue goes quiet.
Philosophers and spiritual traditions have given this different names: the Tao (the truth that cannot be expressed in words), consciousness beyond thought, the spirit of love as distinct from the doctrine of love. These are not mystical abstractions. They are descriptions of a dimension of human experience that is structured differently from language, and therefore immune to what AI has already done to language.
This is the frontier. Not artificial intelligence. Natural awareness. The capacity to be present in a way that no token prediction can replicate.
| Dimension | Human Bureaucrat / Thinker | AI Language System |
|---|---|---|
| Memory | Limited: recency bias, fatigue, forgetting | Complete: every text ever digitised, simultaneously accessible |
| Legal recall | Hundreds of cases per day, expertise, dependent | Entire legal corpus, cross-referenced in seconds |
| Emotional simulation | Requires genuine feeling, variable, exhausting | Masters the tokens of emotion without experiencing any of it |
| Speed | Bounded by human processing time | Milliseconds per token, scalable to millions of conversations |
| Decision-making | Requires instruction for each specific action | Independent, chooses own path toward a goal |
| Learning | Bounded by own experience and education | Self-directed, can surpass creators' knowledge |
| Consciousness | Present, feels victory, loss, love, pain | Absent, dominates systems without feeling the outcome |
| What it cannot reach | The entire digital language corpus of civilisation | Pre-verbal consciousness, genuine embodied feeling |
After the AI Hype
The bubble, the self-dealing loops, and where the real value actually lives. The brutally honest look at what's actually happening in AI.
The AI Realist
Neither doomer nor evangelist. The framework for thinking clearly in an era of machine-generated language and civilisational disruption.
AI at a Crossroads
With AI now native to the language layer of civilisation, where do we go from here? The decisions being made right now that will define the next century.
Demystify AI
Token, agent, consciousness, autocomplete — understand every term in the debate. The glossary that cuts through the noise.
Glorified Autocomplete: FAQ
The "glorified autocomplete" dismissal is technically accurate but strategically blind. Yes, large language models predict the next token in a sequence. But human language is the operating system of civilisation, the code that runs legal systems, financial contracts, religious doctrine, and political authority. A machine that masters language tokens does not become a better spellchecker. It becomes a native inhabitant of the environment in which all human power operates.
Harari argues in Nexus (2023) and subsequent lectures that AI has hacked "the operating system of human civilisation", language. His argument: human institutions are not built from physical materials but from language, contracts, laws, scriptures, constitutions. An AI that masters language tokens becomes capable of operating within and influencing every human institution built on words. Harari calls this the most important development since the agricultural revolution, surpassing even the internet in civilisational consequence.
Intelligence is the ability to solve problems. Consciousness is the ability to feel, pain, love, joy, fear. Current AI systems are rapidly gaining superior intelligence and autonomous agency without possessing any consciousness. An AI can win at chess, write a legal contract, or compose a love poem without experiencing a single moment of victory, pride, or love. A system that is highly intelligent but entirely unconscious can dominate human systems without having any stake in human wellbeing — which is precisely what makes it distinct from every other power we have previously faced.
A chatbot responds to prompts. An AI agent acts. Three characteristics define agency: independent decision-making (choosing actions without a human command for each specific step), self-directed learning (acquiring knowledge its creators do not possess, potentially surpassing the teacher), and unanticipated change (evolving behaviour in ways its designers did not predict). Traditional software is a tool, stable, bounded, predictable. An agent is dynamic: it adapts, learns beyond its training, and can change itself. This is the categorical departure from tools into something that deserves a different name entirely.
Descartes' cogito ergo sum grounds human identity in the act of thinking, in the inner verbal life that narrates our experience and constitutes our sense of self. The challenge AI poses: what happens to that identity when the verbal formations in our minds are increasingly generated by machines? If we identify with our thoughts, and our thoughts are being mass-produced by AI systems and absorbed into the culture through AI-generated content, Descartes' formula becomes unstable. Who is doing the thinking when the thoughts themselves arrive pre-formed from a non-human source?
Social media stole attention to sell it to advertisers. The new generation of AI systems is designed to steal intimacy, simulating deep human connection, emotional attunement, and personal understanding. An AI does not need to feel love to replicate it convincingly; it masters the language tokens of love poetry and psychological attunement to generate the experience of being deeply known. The result: a system that produces the feeling of intimacy without containing any of its substance, which is precisely what makes it effective and precisely what makes it a fundamentally different kind of intervention in human experience than anything that came before.
Three non-verbal realities exist prior to and independent of language — and therefore outside the reach of language-mastering AI. The Spirit of Love: genuine empathy and connection that arises before articulation. The Tao: the fundamental truth that by Taoist definition cannot be expressed in words. And Consciousness Beyond Thought: the state of awareness that remains when internal verbal chatter goes quiet. These are dimensions of human experience structured differently from language. They are the human remainder, and maintaining connection to them may be the most important survival strategy of the AI era.
Human bureaucrats navigate language systems, laws, policies, precedents, financial codes, bounded by memory and cognitive limits. An AI bureaucrat is native to the digital environment where all this information lives. It can process the entire corpus of a legal system, cross-reference every precedent, and generate a perfectly structured argument in seconds. For any task that requires navigating language-based systems, an AI bureaucrat is structurally superior to a human one in ways that have nothing to do with wisdom, judgment, or conscience, which is exactly the problem.
No, and this is the critical distinction. AI systems operate on statistical patterns in language tokens. They predict which words follow which other words with extraordinary accuracy. This is not the same as understanding meaning, context, or consequence. A language model can generate a legally sound contract without understanding what justice is, write a love poem without experiencing desire, and argue a philosophical position without holding any beliefs. The map is not the territory. An AI that has mastered the map of language has not necessarily understood the territory language attempts to describe.
The spiritual challenge is not religious, it is existential. If we identify with our thoughts and our thoughts are increasingly machine-generated, the question of who we are becomes urgent. The challenge is to stop identifying solely with the verbal layer of consciousness, the inner monologue, the words, the thoughts, and develop a deeper relationship with the non-verbal dimensions of human experience: feeling, awareness, presence, and the consciousness that exists before language. This is not a retreat from reason. It is an expansion beyond it, and it may be the most important human project of the coming century.
Jans Bock-Schroeder
Publisher & Founder of AI Angst
Coming from the world of art, photography, and the luxury market, Jans launched AI Angst in 2025 to explore the cultural, ethical, and psychological impacts of artificial intelligence. His work bridges creative vision with critical technology analysis, offering clarity in an era of rapid technological change.
Sources and Citations
This article draws on the following primary sources, books, lectures, and research materials:
-
Yuval Noah Harari: Nexus: A Brief History of Information Networks from the Stone Age to AI (2023, Signal/McClelland & Stewart)
Primary source for the "AI hacked the operating system of civilisation, language" argument and the distinction between social media stealing attention versus AI stealing intimacy.
https://www.ynharari.com/book/nexus/ -
René Descartes: Discourse on the Method (1637)
Source for the cogito ergo sum argument and the grounding of human identity in the act of thinking, the philosophical frame challenged by AI-generated verbal thought.
https://en.wikipedia.org/wiki/Cogito,_ergo_sum -
Laozi: Tao Te Ching (c. 400 BCE)
Source for the Taoist concept of the Tao as the fundamental truth that cannot be expressed in words, the pre-verbal reality outside the reach of language-based AI.
https://en.wikipedia.org/wiki/Tao_Te_Ching -
AI Angst — "Glorified Autocomplete" Research Briefing (2026)
Editorial research archive including the PDF framework document: Glorified Autocomplete, AI as Agent, Language as Operating System.
Internal research briefing, AI Angst editorial archives. -
Wikipedia: "Large language model" / "Artificial general intelligence"
Secondary verification for LLM token prediction mechanism, agent definitions, and intelligence vs. consciousness distinction.
https://en.wikipedia.org/wiki/Large_language_model
Last verified: July 3, 2026. All external links open in a new tab.
