AI in schools does not need to mean cheating, copy-paste answers, or shallow thinking. Here’s how teachers and workplace trainers can use AI to build deeper learning, stronger discussion, and better practice.
AI in schools: how to make AI the thinking partner, not the shortcut
AI in schools has a reputation problem.
Ask a room of teachers what worries them about AI in education and you will hear a fair list. Students using AI to cheat. Debate responses that all sound the same. Essays with no real argument. Classroom bans. Teachers wondering where the student’s thinking ends and the chatbot begins.
Should AI be used in schools?
Fair concerns.
But banning AI from learning is like banning calculators from maths forever. You might delay the problem. You will not prepare learners for the world they are already in.
The better question is not “Should students use AI?”
The better question is “What kind of thinking do we want AI to support?”
That question matters in classrooms. It matters just as much in workplace learning, where staff are already using AI tools to draft emails, build reports, summarise content, write scripts, test ideas, and make decisions faster than the policy team can say, “We need a working group.”
AI is here. Now we need better learning design around it.
What The Deep Learning Revolution teaches us about learning
Terrence J. Sejnowski’s The Deep Learning Revolution tells the story of how deep learning moved from a niche academic field into a major driver of modern AI. MIT Press describes the book as a look at how deep learning shifted AI away from older logic-and-symbol systems and toward neural networks that learn from data.
That shift matters for education.
Deep learning systems do not learn by memorising one neat rule. They improve through exposure, feedback, pattern recognition, adjustment, and repeated testing. MIT Press notes that these systems are fuelled by data and learn in a way that can be compared to how babies experience the world, starting fresh and building capability through repeated contact with new environments.
Human learning is not the same as machine learning. Let’s not get weird about it.
But the overlap is useful.
People also learn through feedback. Through examples. Through trying, getting it wrong, trying again, comparing responses, testing ideas, and applying knowledge in context.
That is where AI can help.
Not by handing learners the answer.
By giving them better material to think with.
The real risk is not AI. It is lazy learning design.
Michael Peart, Managing Partner at Hungry Minds, hears the concern often.
Teachers and leaders worry that students will use AI to avoid thinking. They worry learners will ask for an answer, copy it, submit it, and learn nothing. They also worry AI will flatten thinking, where every student ends up with the same bland response to a debate topic because they used the same tool in the same way.
That risk is real.
But the problem is not the tool. The problem is the task.
If an assessment asks a learner to “write five arguments for and against school uniforms,” AI can do that in seconds. Snooze.
But if the task asks learners to compare AI-generated arguments, identify weak reasoning, add evidence from class content, defend a position, respond to a peer challenge, and revise their thinking after feedback, we are in a very different room.
Same tool. Better task. More thinking.
How AI can support deeper learning in schools
Used well, AI in the classroom can help teachers shift time away from routine content delivery and toward higher-value learning.
That is the sweet spot.
The source article on Dr. Doreen Mayrell’s flipped college algebra model shows this clearly. Students use AI tutors outside class to work through guided notes, direct instruction, early practice, and prerequisite gaps. Then classroom time stays human. Paper. Pencil. Calculators. Questions. Misconceptions. Complex problems. Real discussion.
That is not AI replacing the teacher.
That is AI clearing space for the teacher to do the work only a teacher can do.
The same model works in schools.
A teacher could use AI to help students prepare before a lesson, generate practice questions at different levels, reframe examples around student interests, or create quick feedback loops. Then class time can focus on debate, application, misconceptions, peer discussion, teacher feedback, and real-world transfer.
Less “everyone watch me explain this for 40 minutes.”
More “show me how you are thinking.”
A practical model for using AI in schools
Here is one way to use AI in schools without turning every task into copy-paste soup.
Before class, students use an AI tutor or guided prompt to explore a concept. The teacher gives them a structured resource, a PDF, a reading, a worked example, or a short video. The AI does not replace the resource. It helps students question it.
Students ask AI to explain the concept three ways. They ask for examples. They ask for a quiz. They ask what a beginner might misunderstand. They bring one useful idea and one question to class.
In class, AI goes away.
Students compare answers. They test examples. They identify weak explanations. They correct errors. They apply the concept to a fresh scenario. They build their own argument.
After class, AI comes back as a reflection tool.
Students ask AI to quiz them, help them review feedback, or suggest where their reasoning needs more work. The teacher can ask students to submit the prompt trail, reflection notes, or a short “what I changed after feedback” statement.
That last piece matters.
It makes thinking visible
AI in workplace learning
The same approach works in organisational training.
At Hungry Minds, we have been testing AI as a learning partner, not a content vending machine.
Michael explains one example:
“We recently designed a handling difficult conversations program where participants used ChatGPT to generate easy, medium, and complex scripts. Then learners worked through how they would respond. They discussed the strengths and weaknesses, debated the options, and used ChatGPT to generate more responses. Those responses became material for role-play in the training room. It was a great example of using AI with scaffolding to progress the learning.”
That is the bit many AI conversations miss.
The AI output was not the endpoint.
It was the stimulus.
Learners still had to judge, adapt, respond, practise, and reflect. That is where the learning happened.
For workplace training, AI can support deeper learning by helping learners:
- Generate realistic scenarios for practice
- Compare easy, medium, and complex examples
- Role-play hard conversations
- Test different customer, client, or stakeholder responses
- Identify gaps in their own judgement
- Build personalised practice pathways
- Reflect on what changed after feedback
This works well in leadership, customer service, safety, compliance, onboarding, sales, and technical training. Anywhere people need to make decisions, explain ideas, handle pressure, or adapt in the moment.
Which is to say, most jobs.
Want examples of this kind of work in action? Start with Hungry Minds’ custom eLearning development, learning design, and learning strategy consulting services.
The PDF prompt method
One simple way to introduce AI to reluctant staff is to start with a trusted resource.
Michael has been creating PDF-style summaries of learning courses, then giving people instructions and a prompt to explore the resource however they see fit.
This works because the AI task has guardrails.
Staff are not staring at a blank chatbot box thinking, “What now?”
They have a resource. They have a purpose. They have permission to ask questions in their own way.
For AI-reluctant staff, that matters.
A good starter prompt might look like this:
You are helping me understand this learning resource. Ask me what I want to achieve first. Then help me explore the content in a way that suits my role. Use the resource as your main source. Do not give me a summary unless I ask for one. Ask questions that help me think, apply, and reflect.
That prompt changes the experience.
AI becomes a coach, not a cheat code.
How to stop AI making every answer sound the same
Here is the classroom problem many teachers can spot in three seconds.
Thirty students submit an argument. Every response has the same structure, same tone, same vague examples, and the same oddly polished ending.
The fix is not only detection.
The fix is task design.
Ask learners to include:
- A personal position and why they hold it
- A link to class discussion or a specific source
- A critique of an AI-generated answer
- A change they made after peer or teacher feedback
- A local example
- A counterargument they disagree with
- A short reflection on what AI got wrong or missed
This makes the task harder to outsource.
More importantly, it makes the task worth doing.
The same principle applies in workplace learning. Do not ask staff to use AI to “write a response to a difficult customer.” Ask them to compare three possible responses, identify the risk in each one, rewrite the strongest option in their own voice, role-play it, and reflect on what worked.
AI gives the raw material.
The learner does the thinking.
That is the deal.
Deeper learning in the age of AI
Here is Michael’s take:
“For me, deeper learning in the age of AI means moving past the first answer. AI can generate a neat response in seconds, but that’s not where the learning is. The learning happens when people question it, test it, adapt it, argue with it, and apply it to a real problem. Used well, AI gives us more material to think with. Used poorly, it becomes a shortcut around thinking. That’s the line educators and organisations need to hold.”
That line is the whole game.
AI can support retrieval practice, feedback, self-paced learning, scenario generation, coaching, debate, role-play, reflection, and revision.
But it needs design.
Without design, AI becomes a faster worksheet machine.
With design, it becomes part of a richer learning experience.
Where Hungry Minds fits
At Hungry Minds, we design learning people lean into. That includes AI-supported learning that still respects the role of teachers, facilitators, coaches, peers, and practice.
We use structured models like ADDIE to analyse learner needs, design the right experience, build practical materials, support rollout, and evaluate what worked. You can see more in our instructional design services and AI in learning design guide.
We also know education. Our work with Autism CRC turned detailed autism research into teacher-friendly eLearning used by educators across Australia with case studies, videos, and downloadable resources. Read the Autism CRC case study and our instructional design IRL article for more on that.
If your team needs to build this capability in-house, our Professional Certificate in Instructional Design helps people learn the process, tools, templates, and practical design skills to create learning that does more than fill a calendar slot.
The takeaway
AI in schools and workplace learning is not good or bad on its own.
It depends on the learning design wrapped around it.
Use AI to avoid thinking, and you get shallow work faster.
Use AI to spark questioning, comparison, practice, feedback, and real application, and you get something better.
More thinking.
Less yawning.
A decent result, really.
