Most people are worried about whether they can trust artificial intelligence.
I have a different concern.
I’m starting not to trust the people using artificial intelligence.
Specifically, there is a new generation of AI users—many of them under 30—who have been handed what may be the greatest learning tool in the history of mankind and somehow figured out how to use it to stop thinking.
Think about this.
For thousands of years, if you wanted to know something, you actually had to know something.
You read books.
You studied.
You asked people who knew more than you.
You went to school.
You researched.
You occasionally had to sit quietly for three consecutive minutes and actually think.
Now?
Every 24-year-old on Earth is walking around with Aristotle, Einstein, Warren Buffett, a cardiologist, a constitutional lawyer, a software engineer, a nuclear physicist, a therapist, a chef, and a McKinsey consultant living inside their iPhone.
And somehow we are getting dumber.
It’s remarkable.
I have never seen a generation with the ability to learn almost anything, understand almost anything, and access virtually the entire accumulated knowledge of mankind…
…and then go full-on retard the second you ask them to use their own brain.
Yes, I know.
“Full-on retard” is not politically correct.
Neither am I.
I call it as I see it.
And I have developed a name for this emerging species:
AITards.
Pronounced A-Tards.
AITards don’t use artificial intelligence.
They outsource consciousness to it.
There’s a difference.
A normal person might ask ChatGPT or Claude a question, read the answer, think about it, challenge it, check the assumptions, maybe ask a follow-up question, and ultimately form an opinion.
An AITard asks Claude a question and receives the Ten Commandments.
Conversation over.
I recently questioned something somebody presented to me.
Their response was essentially:
“Claude says you’re wrong.”
Oh.
Well.
My apologies.
I didn’t realize Claude had joined the meeting.
Should I move over?
Would Claude like some coffee?
Does Claude validate parking?
Because apparently Claude is now the final authority on everything.
And these people don’t merely repeat the answer.
They present it with the confidence of someone who personally spent 27 years researching the subject at MIT.
They suddenly become experts.
Finance.
Medicine.
Law.
Artificial intelligence.
Real estate.
Quantum physics.
Agriculture.
Whatever you want.
They asked a chatbot one question at 9:17 Tuesday morning and by 9:19 they apparently had a double PhD.
So I started doing something incredibly unfair.
I asked them a follow-up question.
This is where everything falls apart.
“What does that actually mean?”
Blank stare.
“Why is that assumption correct?”
Longer blank stare.
“What information did you give AI to arrive at that conclusion?”
Now I have apparently asked them to derive Einstein’s field equations using a cocktail napkin.
Finally:
“It’s right there.”
WHAT IS RIGHT THERE?
Explain it to me.
You just presented a 14-page analysis with charts, projections, market assumptions, competitive analysis, and a conclusion delivered with the confidence of Winston Churchill addressing Parliament.
Surely you can explain paragraph two.
“It’s right there.”
No.
The correct answer is:
“I have absolutely no fucking idea.”
And strangely, I would respect that answer.
At least it would be accurate.
But my favorite question is:
“What exactly did you put into AI to get this answer?”
Because AI is extraordinary, but it has one slight limitation:
It generally works with the information you give it.
Garbage in, beautifully formatted garbage out.
And inevitably the answer is something like:
“I put in what you told me.”
Excuse me?
So let me understand this.
I gave you the information.
You typed my information into Claude.
Claude reorganized my information into bullet points.
You brought Claude’s version of my information back to me.
And now you’re telling me I’m wrong because Claude said so?
What in the friggen world is happening?
We’ve somehow invented intellectual money laundering.
My own thoughts go into a computer, come back wearing a blazer and carrying a PowerPoint presentation, and suddenly they’re independent expert research.
This is the part people are missing about AI.
AI can make smart people dramatically smarter.
But it can also make stupid people spectacularly confident.
AI didn’t eliminate bullshit. It industrialized it.
That may turn out to be one of the great unintended consequences of artificial intelligence.
Bullshit used to be hard work.
You had to make it up yourself.
You had to remember it.
You had to defend it.
And eventually, if somebody kept asking questions, the whole thing usually collapsed.
Now?
Bullshit comes professionally formatted.
It has headings.
It has footnotes.
It has charts.
It has an executive summary.
It has 14 citations nobody bothered to read.
Ask whether they actually read one of the citations and suddenly we’re experiencing technical difficulties.
AI can now take an absolutely terrible idea and, in approximately eight seconds, turn it into something that looks like it was prepared by Goldman Sachs.
The bullshit didn’t get smarter. The formatting did.
And that may be the more dangerous development.
Because before AI, there were warning signs.
If someone didn’t understand something, eventually you could tell.
They stumbled.
They hesitated.
They couldn’t explain it.
Now AI gives everyone perfect grammar, beautiful formatting, executive summaries, citations, charts, bullet points, and phrases like:
“Based on a comprehensive analysis of the available data…”
Oh shit.
We have no idea who knows anything anymore.
The person who spent 20 years studying the subject and the person who discovered the subject 11 minutes ago can now produce documents that look remarkably similar.
For years, everyone worried that artificial intelligence would eventually become indistinguishable from humans.
Maybe we worried about the wrong thing.
Maybe we should have been worried that humans would become indistinguishable from AI output.
There is only one remaining test:
Ask a second question.
That is the new Turing Test.
Not whether the computer can convince you it’s human.
Whether the human can convince you they’re not a computer terminal.
Ask:
“Why?”
Then ask:
“How do you know?”
Then:
“What assumptions is that based on?”
And finally:
“Do you actually understand this?”
At that point, you will discover whether you’re talking to a human being who used AI…
…or an AITard who rented a brain for 45 seconds.
For the record, I love AI.
I use it constantly.
I think it may be the most transformational technology of our lifetime.
But the entire point of having the greatest thinking tool ever invented should be to think better.
Not to stop thinking altogether.
AI should make you more curious.
It should make you ask better questions.
It should allow you to challenge assumptions faster.
It should help you learn things that previously would have taken years.
It should expand your brain.
It should not replace it.
So maybe the next great AI company doesn’t need to build a smarter model.
Maybe it just needs one giant button that appears after every answer:
“DO YOU ACTUALLY UNDERSTAND WHAT YOU JUST READ?”
And if you click “Yes,” you have to explain it in your own words.
No Claude.
No ChatGPT.
No phone.
Just you.
That should clear up the problem pretty quickly.
Because I’m not particularly afraid that artificial intelligence is going to take over the world.
I’m increasingly concerned that humans are simply going to hand it the keys.