How to Judge an AI Course for Your Teenager

Quick answer

Judge an AI course for your teenager on seven structural markers: live instruction from a named human, a cohort small enough that your child gets noticed, projects rather than videos, scheduled critique, academic integrity taught explicitly, a showcase at the end, and a pathway that continues after the last session. A course that clears all seven will build real capability. A course that fails three or more is usually a video library with a certificate stapled on. You do not need to understand AI to run this check. Every marker can be verified in one phone call, and the fastest single test is to ask for finished student work from the last cohort.

Why quality is hard to see from the outside

AI education is a young market, and young markets attract two kinds of sellers: people who have thought hard about how teenagers learn, and people who have thought hard about how parents buy. From the outside they look identical. The websites use the same vocabulary, the testimonials glow at the same temperature, and the certificates look equally official.

The pressure to choose something is real. An Elevate Education survey found roughly three-quarters of Australian high-schoolers already use AI at least a few times a week, and PwC's 2025 Global AI Jobs Barometer found that jobs requiring AI skills carry a 56% wage premium. Doing nothing has a cost. Choosing badly has a different one: a bored teenager, a wasted term, and a quiet family lesson that AI courses are a rip-off.

The fix is to stop reading marketing and start checking structure. The seven markers below are structural, which makes them hard to fake and cheap to verify. They sit inside the bigger national picture covered in AI education for teenagers in Australia, but they work as a standalone test.

The seven quality markers

Three of these carry the most weight. Projects, not videos matters because watching explanations produces the feeling of learning without the substance; building something and fixing it when it breaks produces the real thing. Critique matters because feedback on work in progress is how students actually improve; a grade delivered afterwards changes nothing. And the showcase matters because a public deadline changes how a teenager works for the whole course, not just the final week. The case for that last one is made in full in why showcase presentations matter.

Red flags that should end the conversation

Some signals are bad enough that you can stop evaluating and move on.

• No humans on the website. If you cannot find out who teaches, nobody teaches. You are looking at a content library.
• Self-paced as the main selling point. Self-paced means alone, and very few teenagers finish a self-paced course alone.
• Certificates as the headline outcome. A certificate from an unknown provider is worth whatever an admissions officer decides it is worth, which is usually nothing. Finished projects are the evidence that travels.
• Unlimited enrolment. If there is no cap on numbers, there is no plan for your child specifically.
• No position on integrity. A course that teaches teenagers to generate output without teaching disclosure is training exactly the habit that gets them into trouble at school.
• A vague curriculum plus urgency pricing. Master AI in a weekend, 70% off if you enrol today, is a sentence about the seller's cash flow, not your child's education.

How to run the check in one phone call

You do not need a technical background. You need ten minutes and a willingness to ask plain questions. Call or email the provider and work through the list: who teaches this, how many students in a class, what will my child have built by week three, how does feedback work, what is your position on AI and school assessment, what happens at the end, and what comes next.

Listen for specificity. Good programs answer in nouns and numbers; weak ones answer in adjectives. Then finish with the shortcut question above and see whether student work arrives, or a curriculum PDF does. A longer list, with the reasoning behind each question, is in questions to ask before enrolling in an AI program.

What this looks like in practice

For calibration, here is a program deliberately built around these markers. Edison's Generalist AI Bootcamp is the academy's open-entry program for ages 13 to 18: live cohorts of 12 to 16, in Sydney, Melbourne and online, over four or eight weeks. Students build from the first sessions, get structured critique on work in progress, are taught to use AI honestly and disclose it, and finish by presenting their project at a showcase. Students who want more can continue into Edison's selective programs afterwards.

Hold every provider, Edison included, to the same seven markers. If a cheaper course clears all seven, it may genuinely be the better choice for your family. Most will not clear three.

The recommendation: run the seven-marker check before you pay for anything. Ask who teaches, how big the class is, what gets built, how feedback works, how integrity is handled, what happens at the end and what comes next, and always ask to see finished student work. A provider that answers all seven plainly belongs on your shortlist. A provider that dodges two or more has already answered the only question that matters.