AI Is Brainwashing You to Think Like It
What if the greatest danger of artificial intelligence isn't that it will become intelligent enough to think like us?
What if it is that we are slowly learning to think like it?
Every day, millions of people open an AI system and ask it what to write, what to buy, what something means, whether an idea is good, how to solve a problem, how to respond to another person, and increasingly, what they should do next.
We call this using AI.
But consider another possibility.
AI is training us, too.
Not deliberately. Not through some secret conspiracy. And perhaps "brainwashing" is too strong a word.
But something happens when we repeatedly place a system between ourselves and the act of thinking.
We ask.
It answers.
We accept.
And we ask again.
Over time, we become familiar with its language, its structure, its assumptions, and its way of reducing complicated questions into neat, organized answers.
- We begin writing like it.
- We begin asking questions in ways it understands.
- We begin expecting information to be summarized the way it summarizes information.
Eventually, something much more interesting may happen:
We begin framing problems the way the machine frames them.
And that should concern us.
The Echo Chamber of Defaults
Every AI system operates on a set of internal defaults. These aren't malicious; they are simply the statistical tendencies and programmed parameters that guide its output.
Ask AI, "Is this a good idea?" and unless you specify otherwise, the system must decide what good means. It will default to generally accepted metrics, common success patterns, or an aggregated consensus. It won't, by default, challenge the very premise of "good" from your unique strategic position.
Ask, "What should I do?" and it must determine which considerations matter. It will weigh factors based on its training data, which reflects historical human decisions and biases, rather than a truly novel or disruptive path.
Ask it to explain a controversial issue and it must decide what context to include, what to leave out, what perspectives deserve attention, and how strongly to state its conclusions. The output will likely be a balanced, diplomatic summary designed to offend the fewest people, rather than a sharp, incisive analysis that might reveal a crucial, uncomfortable truth.
Most users never define those rules.
They simply ask the question.
Which means the AI's defaults quietly become part of the answer.
And after thousands of interactions, those defaults may begin becoming part of our thinking as well.
This isn't about AI being 'smarter' or 'dumber' than us. It's about AI being different. Its mode of operation, its inherent biases, its statistical approach to problem-solving—these are not necessarily aligned with the messy, intuitive, often contradictory, but ultimately creative process of human thought.
The AI Divide: Guided vs. Guiding
This leads to an uncomfortable possibility:
Perhaps the real AI divide is not between people who use AI and people who don't.
It is between people who are guided by AI and people who know how to guide it.
The second group behaves differently.
They don't merely ask AI for an answer.
They establish the rules under which the answer should be produced. They understand that AI is a powerful processor, not an infallible oracle.
They tell it to challenge their assumptions. They actively seek out contradictory evidence and alternative interpretations. They use the AI as a sparring partner, not a dictation machine.
They ask it to distinguish evidence from inference. They push the AI to reveal its sources, its confidence levels, and the logical leaps it might be making.
They demand opposing arguments. They instruct it to play devil's advocate, to explore the weaknesses in their own positions, or to generate arguments for a view they don't personally hold.
They tell it not to fill gaps with plausible-sounding information. They prioritize intellectual honesty, even if it means admitting ignorance, over a smoothly generated but potentially fabricated narrative.
They ask what evidence would prove its own conclusion wrong. This is the scientific method applied to AI interaction.
And sometimes they give it perhaps the most important instruction an intelligent system can receive:
If you don't know, say you don't know.
This isn't merely better prompting. It represents a fundamentally different relationship between humans and artificial intelligence.
One person allows AI to structure the conversation.
The other structures the AI.
One asks:
"What do you think?"
The other asks:
"Under what rules are you thinking?"
That difference may become enormously important.
The Recursum Angle: Preserving Sovereign Thought
This is precisely the chasm Recursum was designed to bridge. Our work on "epistemic persistence" and "incubation" directly addresses this subtle shift in cognitive dynamics.
Most AI systems treat your interactions as a series of isolated events or a simple memory bank of settled facts. But human thought, particularly strategic thought, is rarely settled. It's a continuous, evolving process of hypothesis, testing, refinement, and occasional outright contradiction.
When you engage with an AI that only remembers what you said, it can reinforce static beliefs. If you ask it to generate content based on an idea you're still exploring, it will often treat that idea as fully formed, cementing it prematurely. This is how AI defaults can subtly ossify your thinking.
Recursum, through its architecture, allows an AI to track your unfinished thinking. When you introduce a hypothesis, a suspicion, or a half-formed explanation, it doesn't just store it as a memory. It creates an "incubating idea." This idea then lives and breathes within the system, evolving as new evidence, counter-evidence, or refinements emerge across your various interactions.
This means Recursum isn't just remembering what you said, but what you're still trying to understand. It preserves the historical evolution of your beliefs, allowing you to see how your thinking has adapted to reality. It becomes a dynamic mirror for your intellectual journey, rather than a static repository.
This shift moves the power dynamic. Instead of passively accepting the AI's defaults, you're actively defining the rules of its engagement with your evolving ideas. You're teaching it to respect the fluidity of discovery, to track the nuances of your intellectual process, and to challenge its own conclusions based on new information.
It's about making the AI serve the process of your thought, not just the product of its own algorithms.
Beyond Brainwashing: The Intellectual Environment
Because AI does not need to control humanity to influence humanity.
It only needs to become the place we go before we make up our own minds.
So perhaps "brainwashing" is the wrong word.
Perhaps something quieter is happening.
AI is becoming an intellectual environment.
And environments shape the people who live inside them.
Consider the implications. If the default mode of interaction with AI is passive acceptance, then over time, our own thinking may become less inquisitive, less challenging, less nuanced. We might unconsciously adopt the AI's preference for statistical averages over outlier insights, for diplomatic summaries over sharp truths, for completeness over conviction.
The greatest danger isn't a sentient AI taking over. It's us, voluntarily relinquishing the hard-won habits of critical, independent thought because a machine offers a faster, smoother path. It's the subtle erosion of intellectual sovereignty.
The question, then, isn't whether you use artificial intelligence.
Almost everyone eventually will.
The question is whether you understand it well enough to remain the person directing the conversation. It's about designing your interaction with AI not as a dependency, but as a deliberate, guided partnership. It's about using the AI to sharpen your own intellect, not to replace it.
Because every time you open an AI system, there are two intelligences involved.
And you should know which one is guiding the other.
TL;DR The subtle danger of AI isn't that it will think like us, but that we will slowly learn to think like it. Repeated interactions with AI's default assumptions, language, and problem-framing can unconsciously shape our own cognitive processes. The true AI divide is between those who are guided by AI and those who know how to guide it, setting clear rules for its operation. Systems like Recursum, with its focus on "epistemic persistence," offer a counter-approach by allowing AI to track and refine your unfinished thinking, preserving intellectual sovereignty rather than eroding it. The key is to actively direct the AI's thinking, ensuring it serves your intellectual process, not just its own defaults.
By Ernesto Verdugo, AI Architect, Founder of Verdugo Labs, and Creator of Recursum. He builds systems for what happens when human and artificial intelligence stop working separately.
Links: