By Daniel Stih
Artificial intelligence is supposed to make us smarter. Cleaner decisions. Better information. Less time wasted. According to a recent study, however, the more people rely on AI for answers, the less capable they become of detecting misinformation on their own. Their “unassisted ability to detect lies” drops. The researchers claim this is a good thing — because, they say, AI reduces belief in misinformation.
The study contains a fatal assumption: It assumes AI is always right.Anyone who has used AI for more than five minutes knows that isn’t true. Every AI platform includes a disclaimer: “AI can make mistakes.”If a tool that makes mistakes is training people to depend on it, the danger isn’t just bad answers - it’s losing the ability to think for yourself. I’ll prove it with a simple example possible; how AI told me to clean mold with vinegar.
The Vinegar Mold Problem
When I asked AI how to clean up mold, here’s what it confidently told me: "Use vinegar. For areas “less than 10 square feet”. Spray undiluted white vinegar. Let it sit for “at least one hour.”
Sounds authoritative.Except every part of that answer was wrong. The “10 square foot rule” wasn’t created by the EPA. It came from a single person in the New York City health department decades ago, and was never supported by data. AI scraped it, repeated it, and presented it as scientific truth.
Second, vinegar is not an effective mold remediation tool. It may discolor mold — the same way a dog peeing on a weed may eventually kill the weed — the mold is still there. The spores can still be viable. You haven’t cleaned anything. What you have done is given surviving microbes a competitive advantage. You’ve cleared the field so something more harmful may grow back instead.
In other words, AI took fragments of truth, assembled them into something that “sounds good”… and delivered misinformation dressed up as expertise. This is exactly why the study’s conclusion — that AI reduces belief in misinformation — needs to be rephrased.
It should say,“AI increases your trust in the system.” The more natural, confident, and authoritative AI sounds, the more likely you are to believe whatever it says — even when it’s wrong.
How AI Quietly Controls What You See
Here’s another example:
I recently recorded an episode on climate change.I used actual temperature data from Phoenix — data showing that temperatures weren’t increasing the way people assumed. AI agreed with my numbers. When I asked it to edit my text for clarity, it added paragraphs I did not write:
- “Global temperature must be homogenized.”
- “You need spatial averaging.”
- “Local data doesn’t matter for climate models.”
When I asked AI why it inserted these claims, it answered, “Accuracy guidelines. I applied an editorial step to prevent accidental misinformation.” In other words, I didn’t ask it for fact. It added them because it believed I should say them. Then it admitted:
“When editing science, health, or climate content, I am instructed to de-emphasize interpretations that contradict mainstream consensus.”
This is not intelligence. This is information curation, which leads to a disturbing question: If AI only gives you one side of a debate, how would you ever learn the other side exists? This is not education. It’s algorithmic herding.
The Danger Isn’t AI Being Wrong — It’s You Stopping Asking Questions
The real threat is not that AI will become super intelligent. The threat is more mundane. AI makes people passive. AI makes people obedient. AI makes people stop thinking. If the system always sounds rational, even when it's wrong, you develop intellectual muscle atrophy. This is no different from overusing a calculator - eventually you forget how to do simple math. You forget how to think.
AI models are trained to avoid dissent, avoid controversy, and avoid “dangerous uncertainty.” This causes people to begin to believe that any idea outside the AI-approved window must be wrong, harmful, or stupid.
This is how smart tools can make smart people dumber - not by replacing your brain - lulling your brain into not being necessary.
Why AI Can Never Replace Critical Thinking
AI is not a neutral entity. Behind every model are programmers, editors, moderation teams, legal teams, corporate incentives, and cultural biases. Someone decides what’s “allowed,” what’s “safe,” what’s “true.”
When people say, “AI will eliminate misinformation,” the honest translation is, “AI will eliminate information that challenges the consensus.” Sometimes the consensus is right; sometimes it’s not. Either way, the algorithmic filter becomes the truth. not reality. The more we trust it blindly, the more intellectually vulnerable we become.
How to Use AI Without Losing Your Mind
There is a way to use AI intelligently. It starts with asking better questions.
1. Break your questions into small pieces.
Don’t ask big, revealing questions like,“Was the moon landing fake?” AI will default to the approved narrative. Instead ask, “Why do some people believe the moon landing was fake?” This bypasses the censorship reflex and gives you real reasoning.
2. Don’t show AI your full hand.
Ask questions in steps. Ask for data, not conclusions. For example, if you ask, “Why must vinegar sit for an hour?” AI will invent that logic. Instead, ask, “Show me data on vinegar vs. mold removal.” AI won’t find any. None exists.
3. Ask “Why?” Again.
You may learn more from its limits than its answers. When AI won’t give you something, ask:
- Why can’t you tell me?
- Who decided that?
- What guideline are you following?
- What counts as misinformation?
4. Go to the library.
Bookshelves don’t censor based on algorithms. Side-by-side reading gives you access to an entire spectrum, including ideas you didn’t know existed. If AI is a flashlight, a library is the sun.
Question Everything — Including AI
Smart tools don’t make you smart. Smart questions do. The minute you stop questioning - whether it’s AI, the government, the media, corporations, or even your own assumptions - you become vulnerable.
I’ve believed bad theories before, not because I’m dumb, rather they sounded good until I asked deeper questions. The cure wasn’t a better answer. The cure was better questions.
AI is a powerful tool — and a dangerous one. It can help you see patterns, clarify ideas, summarize information, and accelerate research. It can also quietly narrow your worldview, not maliciously - mechanically.
Use it. Learn from it. Never surrender your ability to challenge it. If we let smart tools do all the thinking, we won’t just lose access to the truth, we’ll lose the ability to look for it.
This article helps you think clearly in a noisy world, cut through misinformation, and find solutions as applied to using AI, artificial intelligence.