Quick answer
Worksheets were designed for a world where producing the work proved you did the thinking. AI ended that arrangement: a chatbot completes a worksheet invisibly, instantly and rather well. Projects hold up differently. A project produces artefacts - proposals, drafts, decisions, dead ends, a thing that works or doesn't - and finishes with the student explaining and defending what they made. That makes AI use visible and assessable instead of hidden: you can see where the tool helped and whether the student can stand behind the result. For parents comparing programs, this is the test that matters: does the work produce evidence of thinking, or just output?
Why worksheets lost to the chatbot
An Elevate Education survey found roughly three-quarters of Australian high-schoolers use AI at least a few times a week, and almost a quarter use it daily, with ChatGPT the most common tool. Now consider what most homework looks like: summarise the chapter, answer the questions, draft the essay - alone, at home, judged only on the final product. That is precisely the shape of task AI completes best.
The problem is not that students are dishonest. The problem is that the format can no longer distinguish thinking from typing. When the only evidence is the finished product, and the finished product can be generated, the assessment measures access to a chatbot.
Schools know this, which is why assessment is shifting. Parents choosing programs outside school should apply the same scrutiny.
What project-based learning actually is
Project-based learning means students spend weeks building something real - an app, an investigation, a design, a working tool - rather than completing disconnected exercises. Done properly, it is not "posters instead of essays". A genuine project has a real question, real constraints, an audience, and room to fail and iterate.
The learning lives in the decisions. Which approach? Why did the first one break? What did you cut when time ran short? A worksheet has one right answer; a project has a defensible position. That difference is why projects build the judgement that durable, AI-resistant skills are made of - and why they are harder to fake.
How projects make AI use visible
A project accumulates a trail. The proposal, the early sketch, version two after the mentor's critique, the build that failed, the one that worked. Anyone watching that trail develop knows roughly what the student can do - which means AI use stops being a secret and becomes a design choice to discuss in the open.
Good programs make the discussion explicit: what did you use AI for? What did it get wrong? How did you check it? A student who used a chatbot to explore approaches, then made their own decisions, has nothing to hide and much to explain - which is exactly the habit that honest AI referencing in schoolwork builds. A student who pasted the whole thing cannot narrate a single decision. The trail tells you which one you are looking at.
Integrity by design: the artefact plus the defence
The final piece is the defence. Standing in front of an audience, presenting the work, taking questions you did not script. You cannot defend work you did not do; the second unscripted question finds you out. This is why showcase presentations matter far beyond confidence-building - they are integrity infrastructure that needs no detector.
It is also the principle Edison builds on. The Generalist AI Bootcamp ends in a showcase; the AI Hypergeneralist year runs six major projects and closes with a defended capstone. And it aligns with where Australian policy already points: the national framework for generative AI in schools puts transparency among its six guiding principles. Artefact plus defence is transparency, made structural.
What parents should look for in any program
Whatever the provider, the same four questions cut through the brochure.
| Ask the provider | A good answer sounds like |
|---|---|
| "What does my child produce?" | "Projects they keep - working artefacts and a portfolio, not completed exercise sheets." |
| "How do you know the work is theirs?" | "We watch it develop and they defend it - process plus presentation, not detector reports." |
| "Can students use AI?" | "Yes, openly and with disclosure - we teach when and how, rather than pretending it doesn't exist." |
| "What happens at the end?" | "They present to a real audience and take questions." |
If the answers centre on worksheets, completion certificates or AI bans, keep looking. The wider guide to what Australian teens need is in AI education for teenagers in Australia.
The recommendation: judge any program - school or private - by the evidence of thinking it produces. Artefacts that accumulate, AI use that is discussed rather than hidden, and a final defence in front of real people. Worksheets measured effort in a world that no longer exists. Projects measure judgement in the one your child is actually entering.
Frequently asked questions
Related insights
Written by
Alex Scriven
Alex Scriven writes for Edison AI Insights on learning design, assessment and what evidence-based AI education looks like in practice.
Published by Edison AI Academy · About the academy
Learn AI the Edison way, with judgement built in.
Edison AI Academy teaches ambitious Australian students to think, build, and lead with AI through structured, project-based, responsible education.