When nobody believes you did it yourself

Most of the conversation about AI and young people is about jobs: which ones will disappear, which ones will change, which ones are safe. That conversation matters, and we write about it elsewhere on this blog. But as parents, the thing that worries us most is quieter than that.

It is the feeling of earning it.

You know the feeling. You worked on something properly, you got stuck, you got unstuck, and you handed it in knowing every word was yours. Then someone noticed. A teacher wrote "this is really good" in the margin. A manager said "nice job" on your first week. That small moment is how most of us learned that effort is worth it.

For a lot of young people right now, that moment is going missing. Not because they are not doing the work, but because nobody is sure they did.

The detector problem

Teachers have a genuinely hard job. Generative AI arrived in classrooms almost overnight, and many schools have reached for software that promises to tell them whether a piece of writing was produced by a machine.

In the United States, more than 40% of surveyed middle and high school teachers used AI detection tools in the last school year, according to a nationally representative poll by the Center for Democracy and Technology, as reported by NPR in December 2025.

The trouble is that these tools are not very good at their job.

• A 2024 study led by Mike Perkins tested six detectors on 805 pieces of text. Their accuracy was 39.5% before anyone tried to fool them, and fell further when the text was lightly edited. The authors concluded the tools cannot currently be recommended for deciding academic integrity cases. Speaking to NPR, Perkins put it more simply: in his field it is now fairly well established that these tools are "not fit for purpose".
• A Stanford team found that detectors flagged essays by non-native English writers as AI-generated 61.3% of the time on average. Plain, careful writing looks "predictable", and predictable looks like a machine.
• Turnitin, one of the biggest providers, publishes its own figures: about a 4% chance per sentence that a highlighted sentence was actually written by a human. The company itself says its detection should not be the only basis for action against a student.

NPR spoke to a teacher who uploaded a chapter of her own PhD dissertation to one detector. It came back as roughly 89% or 91% AI-written. Every word was hers.

What it does to a student who did the work

The same NPR report followed a 17-year-old who was accused of using AI three times in one term. She had written the work herself. She now runs every piece of homework through several detectors before handing it in and rewrites her own sentences if the software does not like them. That adds about half an hour to every assignment.

Think about what that teaches. Not "write well". Not "think hard". It teaches: your own work is suspicious until a machine says otherwise.

These are US numbers, and UK schools are not identical. But the pressures are the same here. Teachers in England are using AI too, mostly to save time: in Teacher Tapp's polling, 54% have used it for lesson planning and 46% for marking. We are not criticising that. Teachers are overstretched and these tools can genuinely help. The point is that the classroom has changed on both sides of the desk, and the old signals of "you did this, and it was good" have got muddy.

There is a relationship cost as well. In the Center for Democracy and Technology's 2025 survey, half of US students agreed that using AI in class makes them feel less connected to their teacher.

Why the missing "well done" matters

Psychologists have studied where confidence comes from for decades. Albert Bandura's work on self-efficacy, the belief that you can do something, identified four sources. A major review of the research in schools found that the strongest is usually mastery experience: actually doing something hard and succeeding at it.

But mastery only builds confidence if it registers as yours. If your best work is met with suspicion, or with silence, the experience does not land. The effort happened, the reward did not.

We think this is one of the least discussed costs of the AI era for young people. A student who refuses shortcuts, writes every word, and still gets no recognition, or gets accused, can easily draw the wrong conclusion: that effort does not count any more. Some will decide that if nobody believes them anyway, they might as well use the shortcut. That would be a terrible lesson to learn at sixteen.

Speaking as parents

This post comes from a family conversation within the team behind Jobsee. We watched someone we love work incredibly hard on something, do it entirely on her own, and get nothing back but a question mark. No "well done". Just doubt. And we watched what that did to her motivation.

We are not going to tell her story in more detail than that. It is hers. But we suspect a lot of families will recognise it.

What would help

None of this is a reason to ban technology or to blame teachers. Some practical things do help:

• Assess the process, not just the product. Drafts, notes, a quick conversation about the work. These are hard to fake and they put the human back in the loop.
• Treat detector scores as a prompt for a conversation, never as a verdict. That is what the better tools themselves recommend.
• Name effort out loud. Specific feedback ("the way you structured the second argument was strong") tells a young person their thinking was seen.
• Value things that cannot be generated. Turning up, helping someone, finishing a shift, fixing something with your hands.

What Jobsee is doing about it

Jobsee is a jobs platform, not a school, but the same principle runs through everything we are building: we help young people show their own work, we do not replace it.

• The Jobsee application is built from what you have actually done: the team you play for, the family member you look after, the side hustle you run. We help you describe it in the words employers use. We do not invent anything.
• If you want help with a cover letter, Jobsee will only arrange your own answers, in your own voice, and you check it before anything is sent.
• In the future we want to go further, with small, human-verified practical tasks and employer endorsements that say, plainly, "this person did this".

Because the young people we built Jobsee for deserve to hear it: you did this yourself, and it was good.