If AI Can Do the Assignment, What Should School Ask Students to Do? An Asymmetrical Though For Consideration.
Caveat: Given what is happening with AI, is there going to be a seismic shift in schools, given that we are already starting to see the friction of today’s AI capabilities when it comes to education. That tension prompted the question behind this article, “Are we now moving to the point where schools need to start thinking even more asymmetrically about integrating people skills, projects, entrepreneurialism, creativity, and judgment, into education in preparing students for the onlife world of tomorrow."
A student submits a clear, persuasive essay. They can explain the main argument, but when their teacher asks why they chose one piece of evidence over another, they hesitate, what should the teacher make of that? Maybe the student understood the argument but struggled to explain it on the spot, or maybe they accepted an AI suggestion without checking it. What could the teacher ask next to find out?
That classroom exchange gets at a bigger question, “Are we on the verge of a seismic shift in education? If so, where would we see it first? In the tools students use, in the way teachers assess learning, in what schools decide is worth spending time on, or all three?”
What does a finished assignment tell us?
Schools use essays, lab reports, pages of equations, and tests to infer what students understand. If a student can produce a polished answer with very little practice, how much can the finished work tell us on its own?
A 2025 field experiment involving nearly a thousand high school math students offers one example. Students with access to a general purpose AI chatbot performed better during practice, then scored lower on an assessment without AI than students who had practised without it, something commonly known as cognitive offloading. A version designed to guide students with hints largely avoided that drop (1).
In our article, “Is AI Helping Students Think, or Thinking for Them?”, we shared a different example (2). A high school student told us she used AI to rework her own class notes into an explanation she could follow. In her case, the tool acted as a tutor, helping her get to the point where she could engage with the material.
What separates that experience from asking AI to solve the problem outright? If the goal is to learn how to solve a problem independently, what decision does the student need to practise making? If an AI explanation helps them get there, could they show what they understood by tackling a slightly different problem on their own?
What would it mean for schools to think asymmetrically?
Imagine a class asked to reduce food waste in its school cafeteria. An AI tool can suggest a dozen ideas in seconds. Which idea should the students pursue? How would they know?
They might begin by weighing what gets thrown away and talking to cafeteria staff. Some students may say the portions are too big. Staff may point out that students at the back of the lunch line barely have time to eat. Now the team has competing explanations, a limited budget, and one week to try something. Do they test smaller portions, change the lunch lineup, or gather more evidence first?
In our article, “The Urgency of Onlife Critical Thinking Education in the Age of AI,” we argued that young people need practice checking the information they encounter online (3). The cafeteria project gives that skill a purpose. Suppose AI recommends smaller portions. What evidence supports that recommendation? Would evidence from another school apply here? Whose experience would the team need to hear before acting?
The project makes room for skills that can be hard to see in a finished assignment. Students practise listening when cafeteria staff challenge their assumptions. They use creativity when their first idea costs too much. They think like entrepreneurs when they spot a need, test an idea with the people it affects, and adapt it to the resources they have. They exercise judgment when the evidence is incomplete and they still have to choose.
Which of those skills would a polished presentation reveal? What else might we learn by asking each student, “What did you initially believe, and what changed your mind?”
What knowledge would students still need?
Suppose the class runs its trial and food waste falls on Friday. Can students claim their idea worked? What if fewer students ate in the cafeteria that day? What if the menu changed?
Students need enough math to recognize why comparing total waste might mislead them. They need to know how to measure waste per meal and why even that comparison might leave questions open. They need enough command of language to explain to the principal what they found and how confident they are. A project gives them a chance to use those foundations when the problem arrives without a chapter heading.
Which skills should students practise independently until they can use them with confidence? Where might AI give them room to attempt something more demanding? And how often does every student get to practise listening, negotiating, and revising a decision during the school day?
Who gets to learn how to use AI well?
In our article, “The Emerging AI Literacy Divide,” we asked what happens when some students receive structured guidance on using AI and others are left to figure it out (4). Consider two classrooms. In one, every student has an AI account. In the other, the teacher shows the class a generated answer and asks students to find the weakest claim, check it against a source, and explain what they would revise. What would we need to see from students in each classroom before deciding who had learned more?.
That question matters when schools decide what to invest in. If access varies by school or household, what opportunities can schools build into class time so every student gets to practise questioning an AI answer, working through a disagreement, and defending a decision with evidence?
How would we know if education had really changed?
A school could put AI in every classroom and still grade the same assignments in much the same way. A deeper shift might be visible in the work teachers ask students to show. A first attempt could reveal where a student gets stuck. A team discussion could surface an assumption no one had checked. A short conversation after a project could show how a student responds when someone challenges their reasoning (2).
Think back to the student who hesitated when asked why they chose a piece of evidence. If students come to believe that their teacher is mainly looking to catch them using AI, a question about their reasoning can start to feel like a test of whether they’ll be believed. “What have I learned, and can I show what I understand?” can give way to “Will my teacher think I used AI, even if I didn’t?” When students feel they have to guard against suspicion, it can become harder to admit where they got stuck, explain why they changed their minds, or ask for help. Those are the moments that can help a teacher see how students are learning to think for themselves.
Think of one assignment students will do next week. If AI could produce a convincing submission for it tonight, what would you ask students to do tomorrow so you could see what they understand, how they work with others, and what they do when their first answer turns out to be wrong? How would you make the purpose of that conversation clear to students?
As AI becomes more woven into everyday life, what will teachers need to help students use it with judgment? We’ve already seen educators rethink where AI fits in a lesson. A teacher might invite students to use it to generate competing explanations, then ask them to decide which one the evidence supports. In another lesson, that same teacher might set AI aside so students can practise building an argument from scratch.
Making those choices takes time, training, and opportunities to compare notes with other educators. Teachers need room to try a pedagogical approach, look at what students actually learned, and revise the lesson from there. If we want students to ask when AI helps them think more carefully, and when it lets them skip a step they need to practise, what opportunities are we giving teachers to work through those same questions?
Digital Food For Thought
The White Hatter
Facts Not Fear, Facts Not Emotions, Enlighten Not Frighten, Know Tech Not No Tech
References
