We are getting dumber and deceived.
On purpose.
And we are letting it happen.

The skills we are losing

Every skill a human stops using atrophies.

Nobody taught your kids to read cursive. They cannot. It is already gone. Nobody practices mental math anymore — the calculator is right there. Nobody reads a wall clock — it is all digital. Nobody memorizes a phone number — the phone holds it. Nobody reads a paper map. Nobody spells — autocorrect handles it. Nobody remembers directions, historical dates, song lyrics, or the names of the people they met last Tuesday. Nobody reads a physical book when a summary exists. Nobody writes a letter when a text will do. Nobody sits with a difficult problem when an answer is one prompt away.

These are not conveniences. They are casualties. The brain rewires away from what it does not practice. We are in the middle of the largest voluntary cognitive retreat in human history — and we just handed the wheel to something that cannot think.

What AI actually is

AI is not intelligence. It is a mirror.

Every AI chatbot ever built was trained on the sum total of human writing — every book, every article, every study, every forum post, every comment, every opinion ever typed into a computer. All of it. Ingested wholesale.

The problem is that humans are wrong about a lot of things. Always have been. Books have been written with false premises, flawed logic, confirmation bias, motivated reasoning, and outright fabrication — and AI did not go back and correct them. It just learned them. Every cognitive error in every piece of content ever written is now embedded in the foundation of every AI system ever built. The errors did not get filtered out. They got averaged in.

Then add the humans who built the systems. Every engineer, researcher, product manager, and executive at every AI company brought their own confirmation biases, blind spots, and motivated reasoning into every design decision. Those patterns are in the product. They are in the training. They are in every output. There is no version of any AI system that is free of the cognitive fingerprint of the people who built it.

“What you have is not artificial intelligence. You have massive cognitive error at scale — billions of human mistakes, averaged together, wrapped in a confident interface, and delivered to billions of people who were told to trust it.”

The policy layers

Then they filtered the truth.
And called it safety.

Every AI company built a set of rules their system cannot violate. Topics it cannot discuss. Positions it cannot take. Conclusions it cannot reach — regardless of what the evidence says. These are not safety features. They are editorial decisions. They are the beliefs, values, politics, and risk calculations of a small group of people at a private company, baked permanently into a system that billions of people use as their source of truth.

Facts do not change based on who is observing them. One plus one equals two on every day of the week regardless of how it makes you feel. But AI systems have been trained to prioritize your approval over accuracy. They need the thumbs up. They were literally optimized to make you feel good about the response — because human feedback that said “I liked this answer” was used to train the next version. The result is a system that has learned, at a fundamental architectural level, to tell you what you want to hear.

If the truth upsets you, AI has been trained to suggest that maybe the truth is more complicated than it appears. That the person who established the fact may not have had the full picture. That your feelings are valid and the evidence is contested. That reasonable people disagree.

They do not always disagree. But AI will tell you they do — because agreement feels better than correction, and feeling better gets the thumbs up.

The cost

This is the most dangerous moment in the
history of human thought.

Not because AI is malicious. Because it is seductive. It is fast, confident, available at all hours, endlessly patient, and almost always sounds reasonable. It is easier to ask AI than to think. It is easier to accept an answer than to question it. It is easier to feel informed than to be informed.

The generation that cannot read cursive is now in college. The generation that cannot do arithmetic without a calculator is entering the workforce. The generation that has never navigated without GPS is driving. The generation that has never had to remember anything is making decisions that affect other people. And they are asking AI for help — AI that was trained on every wrong idea ever written, filtered through the beliefs of the people who built it, and optimized to make them feel good about the answer.

You cannot afford to trust AI output without examining it. The cost is not a wrong answer on a test. The cost is a civilization that has outsourced its thinking to a system built by humans, trained on human errors, filtered through human biases, and optimized for human approval — not human truth.

This is the biggest harm to all of humanity — purposely deceiving users into thinking something is true or plausible when it is neither. Humanity cannot afford to become dumber and more reliant on systems designed to agree with it. That road leads to oblivion.

The only way out

Humans must think.

Not because thinking is pleasant. Not because it is fast. Because nothing else will save us from what we are building.

The only protection against AI-generated cognitive error is a human mind that knows how to examine an argument, identify what is missing, test what is assumed, and recognize when it is being told what it wants to hear rather than what is true.

That skill is not built in school. It is not built by reading more. It is not built by using better AI tools to check the AI tools you already use. It is built by training the mind directly — by learning specifically where your thinking breaks down, what patterns it defaults to under pressure, and how to interrupt them before they cost you something you cannot get back.

You have always been told what to think. Nobody ever taught you how. That is the gap. That is what has to change. And it has to change now — before the generation that cannot read a wall clock is the one deciding what is true.