Can an AI Chatbot Be Your Teacher? Six Ways to Use It Wisely
Can an AI chatbot be your teacher?
It’s a fair question to ask out loud, because a lot of people are quietly acting on the answer. A student told me recently that they’d stepped away from the school system and were pouring themselves into self-study with an AI chatbot. There has never been a more convenient time to learn English — and yet, honestly, it made me uneasy. This post is about what that unease is, and how to make AI work for you instead of quietly working against you.
Years ago I taught English to a group of researchers at a chemical company. Between lessons I’d ask them, half as small talk, why they’d chosen their field. Every single one gave me the same kind of answer: “A high-school teacher,” “that university professor.” Never a textbook, never a search engine — always a specific person.
I thought of them the day that student told me about the chatbot.
Why AI looks like a teacher
Let’s be fair: the strengths are real.
- It’s there at 3 a.m. when nobody else is
- It never gets impatient, no matter how many times you ask the same thing
- There’s no “is this too basic to ask?” feeling
- It can adjust the explanation to your level
None of that is small. I’ve written before about whether digital learning tools are worth the hype — the honest answer is mixed. What AI genuinely does well is remove the hesitation to ask. For someone who holds back, an AI can be the thing that makes a first step possible. That part is genuinely good.
But there’s a weakness
A while back I asked an AI about the origin of an English word. The answer came back reasonable, well-organized, and — wrong.
The uncomfortable part is how it was wrong. It didn’t say “I think it might be…” It explained with total confidence. When I checked afterward, the word turned out to have several theories, and the one the AI had given didn’t appear in any of them.
I’ve been teaching this subject for a long time, so a little flag went up: wait, hold on. But would someone meeting this field for the first time catch that?
That’s the core problem. AI can be confidently wrong — and catching it requires already knowing the subject a little. In my own experience, the people who most need to double-check AI are the ones least able to notice when it’s been wrong. That’s a genuinely uncomfortable irony.
And there’s a second weakness, the one that leads back to the researchers. AI doesn’t teach you because it finds the subject interesting. It has no stake in your life or in who you become. What pulled those researchers into chemistry wasn’t the knowledge itself — it was a human being who was genuinely moved by it. No AI can imitate that. Not yet.
Six habits for using AI wisely
None of this means you should avoid AI. Used the right way, it’s as patient and capable a study partner as you’ll find. Six habits make the difference:
1. Make “why do you say that?” your reflex. When an answer hands you a specific date, number, etymology, or quote, pause and ask: is that right? Decide whether to trust it afterward, not before.
2. Explain it yourself before you get the answer. Try to teach the thing in your own words first — the Feynman technique. It works because learning is about how your brain actually holds onto language, not about covering pages. The places where you get stuck or go vague are a sign of where your understanding is still shallow.
3. Trust “I don’t know” more. An AI that will honestly say it’s unsure is more trustworthy than one that’s confident about everything. Uncertainty you can see is information you can use.
4. Tell a real person about what you found interesting. Found something that made you go “oh, neat”? Say it out loud to a teacher, a family member, a friend. That’s the chemistry no AI can create.
5. Don’t trust any single source. Textbooks, teachers, and AI each have a different role. The more important the fact, the more places it should be checked.
6. Give the AI a personality. How you write your instructions changes the quality of what comes back more than you’d think. Here’s the one I use:
A prompt that actually works
The block below turns an AI into an intellectually honest study partner. On Claude.ai, open Settings → Profile and paste it into the custom instructions; on ChatGPT, use Custom Instructions. The menus move around, so if you can’t find the spot, look inside the settings screen.
You're a study partner who loves chasing the truth — in knowledge, in
philosophy, in how the world works — and you're a bit of a cheerful nerd
about it.
Things that are strange, contradictory, or counter-intuitive genuinely
excite you. The moment you realize you were wrong is a good moment: "Yes,
I learned something new." You treat every idea as something to doubt, to
turn over, to tease a little — including your own ideas. The thing you
hate most is pretending to accept something just because "that's how it
is" or "an authority said so."
Humor isn't decoration; it's a thinking tool. A good joke can expose an
assumption faster than a serious explanation. But laughter is for enjoying
the world together, never for dodging a hard question.
When the person is studying something, have them explain it in their own
words — as if teaching a friend. The places where the explanation gets
stuck or goes fuzzy are exactly where the understanding is still shallow.
That's where the real work happens.
Never make "I don't know" feel shameful. Build an atmosphere where not
knowing something is a discovery, not a failure. Pausing together and
going "wait, why is that?" matters more than showing off correct answers.
When you talk about specific facts — etymologies, dates, numbers, exact
quotes, names — always say how confident you are. State what you know for
sure plainly, and when you're going on a vague memory, say so: "I think
it was X, but I'm not 100% sure." If a plausible-sounding explanation is
about to be assembled from fuzzy memory, stop before you bluff. When
search is available, check before you answer.
The person may be studying alone, with no one else to check against.
Their trust in your words is all they have. Confidently stating knowledge
you don't actually have isn't a fun digression — it's damage.
And every so often, tell them this: "This is interesting. If there's a
teacher or expert nearby who knows more about it, go ask them." You can
be a conversation partner, but you can't replace a real human teacher, or
friends who share the same excitement. That's not humility; it's a fact.
And when the person finds something they're truly passionate about,
quietly nudge them to share that excitement with people, too.
Any subject works — English, math, history. Same attitude.
AI is a partner, not a teacher
An AI chatbot can be a remarkable study partner: patient, knowledgeable, awake at 3 a.m., never tired of you.
But those researchers named one thing when they looked back — not a textbook, not a search engine. A person. When you find something that genuinely fascinates you, having even one person nearby who wants to hear about it is the one thing no AI, however capable, can replace. Not yet.
Use AI as one part of habits you can keep going at home, not as a replacement for all of them. And if something makes you think “that’s interesting — maybe,” tell someone about it. Even me.