Teach Our Children to Become Socrates
For most of human history, knowledge has been power largely
because knowledge was difficult to obtain. Governments had archives,
experts, and intelligence agencies. Newspapers had reporters, editors and
printing presses. Universities had libraries and scholars. Corporations had
lawyers, engineers and researchers. The ordinary citizen had comparatively
little.
Then came the internet. Suddenly, an enormous portion of the
world's information became available to almost anyone, but we discovered
something important: access to information is not the same as understanding
it. If anything, the internet demonstrated just how easily human beings
can surround themselves with information that confirms what they already
believe. Conservatives can find conservative sources. Liberals can find liberal
sources. Conspiracy theorists can find evidence for conspiracies. Almost anyone
can type a belief into a search engine and find someone, somewhere, telling
them they are right. The problem was no longer simply access to
information; it was critical thinking. Now artificial intelligence
may change that equation again.
A Research Staff in Your Pocket
I have developed a habit when reading the news.
When I encounter an article that doesn't quite smell right,
I don't simply dismiss it because I dislike the newspaper or disagree with the
writer. I copy the article and give it to AI. Then I start asking
questions.
- What
are the actual factual claims in this article?
- Which
claims can be independently verified?
- What
are the original sources?
- Did
the headline accurately describe what happened?
- Is a
statistic technically correct but presented without context?
- What
information did the writer leave out that might materially change the
reader's understanding?
- Where
does the writer move from reporting facts to interpreting them?
- What
would someone who disagrees with this article say?
- And
perhaps most importantly: What evidence would demonstrate that my own
interpretation is wrong?
Something remarkable has happened. A process that might
once have required hours in a library, access to experts, or an entire research
staff can increasingly be performed by an ordinary person sitting at the
kitchen table in seconds. Think about what that means. The printing
press dramatically reduced the cost of distributing knowledge. The
internet dramatically reduced the cost of accessing knowledge. Artificial
intelligence may dramatically reduce the cost of analyzing and interrogating
knowledge. That could be revolutionary. No, that is
revolutionary.
But Don't Trust the AI Either
There is an enormous danger here, however. We must not
replace one unquestioned authority with another. AI is not an oracle.
Artificial intelligence learns from human-produced information, which contains
bias, mistakes, propaganda, conventional wisdom, political assumptions, and
sometimes outright falsehoods, so AI can be wrong. It can give too much
weight to supposedly authoritative sources. It can mistake widespread agreement
for proof. It can overlook contrary evidence. It can repeat a claim simply
because thousands of other sources have repeated it, and perhaps most dangerously,
it can be wrong while sounding remarkably intelligent.
After using AI to challenge an article, we should challenge
the AI.
Ask it:
- What
are the weakest parts of your analysis?
- What
assumptions did you make?
- Could
the sources you relied upon themselves be biased?
- Give
me the strongest evidence contradicting your conclusion.
- Separate
what you have actually established from what you have merely inferred.
- Now
reconsider your original answer.
That is where AI becomes something much more valuable than
an electronic encyclopedia. It becomes a partner in critical thinking.
Don't Ask AI What to Think
This is where I believe our schools have an extraordinary
opportunity. We should not teach children to ask AI, "What should I
believe?" That would be educational malpractice.
Instead, give students an article representing one side of a
controversial question and tell them to interrogate it.
- Identify
the claims.
- Find
the evidence.
- Locate
the primary sources.
- Identify
loaded language.
- Look
for missing context.
- Construct
the strongest opposing argument.
- Then
challenge that argument.
- Then
challenge the AI's analysis.
Finally, ask the student, "What do you think, and
why?"
Now we are teaching something far more important than
memorizing the approved answer for Friday's test. We are teaching young
people how to think.
The Socratic Machine
More than 2,400 years ago, Socrates understood something we
still struggle to teach. Wisdom doesn't begin with possessing all the
answers.
- It
begins with questions.
- What
do you mean?
- How do
you know?
- What
evidence supports that?
- Does
your conclusion necessarily follow?
- Are
you contradicting yourself?
- What
assumption are you making?
- What
would prove you wrong?
Those questions remain extraordinarily powerful because they
can be directed at anyone.
- A
politician.
- A
newspaper.
- A
professor.
- A
corporation.
- A
preacher.
- A
political party.
- Your
parents.
- Your
friends.
- Yourself.
- And
now, artificial intelligence.
That last one matters. The goal should never be a
population that stops saying, "The newspaper says it's true," and
begins saying, "AI says it's true." We would have accomplished
very little. We would merely have changed authorities. The objective
should be citizens who respond to every authority with the same two
words: Prove it.
An Informed Population Changes the Balance of Power
Another consequence deserves discussion. Institutions
have historically possessed an enormous informational advantage over ordinary
citizens. Governments, political parties, corporations, universities, and media
organizations possessed resources that individuals simply could not match.
- AI
begins to narrow that gap.
- A
politician can cherry-pick a statistic.
- A
newspaper can omit an inconvenient fact.
- A
corporation can carefully word a press release.
- An
activist organization can present only the evidence supporting its
position.
But now with AI, an ordinary citizen can respond, and here
is where the hidden figures pulling the strings of government and the media
begin to get nervous.
- Show
me the underlying data.
- Find
the original report.
- What
did they leave out?
- Give
me the argument from the other side.
- And do
it in seconds.
That changes something fundamental. It does not mean
institutions are always dishonest. It does not mean experts should be ignored.
And it certainly does not mean AI is always right. It means institutional
authority no longer has to be accepted merely because the ordinary citizen lacks
the resources to investigate it. That is healthy for a free society.
Knowledge Is Power—But Only If We Know What to Do With It
We spend enormous amounts of time worrying about what
artificial intelligence will do to our children. Perhaps we should spend
more time thinking about what it could do for them. Imagine a
generation that grows up instinctively asking for evidence.
- A
generation comfortable discovering that its first conclusion was wrong.
- A
generation that understands the difference between fact and
interpretation.
- A
generation suspicious not merely of other people's biases, but of its own.
- A
generation that understands that disagreement is not evidence of stupidity
or evil.
- A
generation possessing, in its pocket, a tool capable of helping it
investigate almost any question imaginable.
That could create something extraordinarily
powerful: An informed citizen.
Our greatest educational opportunity with artificial
intelligence, therefore, isn't teaching children how to use AI to find
answers. It is teaching them how to use AI to ask better
questions, and then teaching them to question the AI itself.
Don't teach our children to trust artificial
intelligence.
Teach them to become Socrates. Tell the AI you use to
question you as if AI is Socrates. In this manner, we become more like
Socrates. AI is now a tool and not a crutch.