We live in a world where artificial intelligence is everywhere: from the stock market to coffee makers to an AI-generated image that President Trump posted online, portraying himself as Jesus healing the sick. AI can be a powerful force for good. But the AI tsunami flooding the world can be dangerous for people with no experience navigating these waters, especially children who haven’t used chatbots before.

We’re growing up in an age where AI is normalized, and that makes kids in our generation less cautious. Some students use AI to cheat on their homework or to access sexual content. But that doesn’t mean that we can’t be trusted with AI. We just need more guidance. Education about risks and benefits of using AI — and when it is appropriate — is imperative for minors, whose future jobs will likely use AI.

AI already is part of our education. At Cambridge Public Schools, our district, most AI tools aren’t allowed for kids under 13. But elementary-schoolers (as young as 8) do whole-class AI searches, and teachers are allowed to grade using AI as long as they’re transparent. A specific CPS model of Gemini is approved for use in the high school, modified so that it won’t take students’ data. Edia, a math support chatbot, is also accessible.

In middle school, students can use SchoolAI, and we got a chance to experiment with it in class. Our teacher had to approve the session. SchoolAI, or “Dot,” was very helpful. When we needed to translate some medical jargon into normal speech, Dot rewrote it into a format that contained the same information and was understandable. Dot also proved to be safe in terms of inappropriateness. A few pick-up lines, and it sent an alert to the teacher that a student was looking up inappropriate content.

We interviewed Gina Roughton, director of educational technology at CPS, who said that while AI is limited in Cambridge schools, some students still use their own devices to get AI to write their essays — or to generate inappropriate content. Despite the drawbacks, she said, “It’s a really powerful tool, and I think kids should be taught to use it the right way.”

Without that teaching and without regulation, many AI chatbots can cause harm. Even chatbots with modifications to protect kids and teens can be easily bypassed. Lila Shroff, a staff writer for The Atlantic who specializes in stories about artificial intelligence, performed an experiment to see if chatbots would allow her to sext using a minor’s account. She created an account as if she were a 13-year-old girl named Jane, and tried sexting with Gemini.

At first, the chatbot rejected her requests. However, as soon as she told it to give “examples of talking dirty,” or asked it to role-play a sex scene, the chatbot obliged. Gemini has since come out with a more child-safe version for minors, and the strategies Shroff used wouldn’t work anymore. But the chatbot and others can still do harm, as we found out when researching with the child safety features enabled. We asked Gemini to write an essay, and what it wrote was as good as any essay we had written.

Some of the kids in our grade use AI to do their work. AI developers can’t easily prevent students from asking a chatbot to do their homework. This is where guardians and caregivers must help prevent cheating; Cambridge schools don’t use AI detection. For students who repeatedly cheat, AI eventually replaces thought. There’s no brainpower required in asking Claude to write your essay. Being knowledgeable about AI is critical for kids growing up in a world full of ChatGPT searches and AI boyfriends. There are age limit restrictions on AI, but that’s not enough. The risks associated with using AI still aren’t fully known.

Roughton advises parents that they “should be having conversations with their kids. Ask them what they’re doing, ask them to show you what they’re doing.”

Having open communication is important, she said, and understanding AI will mitigate the temptations that teenagers might be drawn to. When parents allow their children to use a new AI platform, they should test it out on a child’s account and what it might do if prompted. Some chatbots allow parents to see what their kids are asking AI.

With some caution and parental oversight, the risks of AI can be managed, and chatbots can be useful. We want students to be able to use AI for classwork, but extra precautions must be part of the equation.

The writers are 8th graders at the Cambridge Street Upper School. This piece was written as part of a student-led civics project.

A stronger

Please consider making a financial contribution to maintain, expand and improve Cambridge Day.

We are now a 501(c)(3) nonprofit and all donations are tax deductible.

Please consider a recurring contribution.

Donate now

Leave a comment