Scroll Smarter: AI hallucinations

Aug. 11, 2026

in this issue

Family Q&A | Top story | Bonus reads | Video of the month | Reader poll

Explaining AI hallucinations to kids 

Question: 

How do we as parents help kids understand how AI “hallucinates” answers? And how can we show them why using AI as a shortcut to homework can lead to incorrect information and lower scores?

— Tired mom

Answer: 

This is one of the most important questions to be asking when it comes to kids’ AI literacy: How can they know when to trust the information they get from AI chatbots?

First, it’s vital that kids understand how AI chatbots work: they don’t retrieve information the way a search engine does; they generate responses by making millions of context-dependent predictions about which words to use in what order. This makes them remarkably good at producing language that feels confident and natural — but it also makes these tools vulnerable to filling in the gaps with language they deem plausible. That’s what has come to be known as “hallucinations,” bad guesses about what words come next.

Students who mistakenly believe that AI models are retrieving information rather than generating it might underestimate the problem of hallucinations. So a good place to start is to ask them to explain how they think AI chatbots work — and why it is that they sometimes get things so completely wrong. (AI models are also prone to flattering us, so they often build on rather than correct things we get wrong.)

From there, it’s a good idea to be sure your child is familiar with examples of AI fabrications that have made their way into the real world. These include legal filings that cited nonexistent cases or other AI errors (more than 1,400 in the last three years) and a refund policy misinterpreted by an airline’s own chatbot (which the company was then legally forced to honor).

Helping students understand the shortcomings of AI can give them a clearer picture of the risks in using it for homework and school assignments. Over half of teens say they have used AI tools for schoolwork, according to a study by Pew Research Center. Have a conversation with your child about their own experience in school. Are they or their classmates using AI tools? Know what they’re using and why.

Finally, you can encourage your child to chat with a bot about topics they know a lot about to see whether they experience false outputs firsthand. Teachers who use this approach report that students learn to rely more on verification techniques than their own intuition to detect AI fabrications.

➡️ Do you have a question for us? Share it in this form and we may answer it in a future issue. We can address questions about artificial intelligence, social media, journalism, misinformation and more in a family-oriented way.

⭐ You’ll win swag from the News Literacy Project if your question is picked! Questions may be edited or rephrased for clarity and length.

Kids turn to chatbots in search of mental health support

More kids are turning to chatbots when they’re sad or anxious, sparking concern among parents. But experts say it’s best to respond with curiosity rather than confrontation or panic.

What to know:

🤖 Nearly one in five teens seek emotional support from AI chatbots. Most kids age 9 to 17 (78%) have used ChatGPT and nearly a third seek advice from AI about romantic feelings or relationships (31%).

🚩 Psychologists say replacing human relationships with digital tool interactions has a negative effect on kids’ emotional development.

What you can do:

📱Ask your kids what AI tools they use and for what purposes. Look at the tools together and help your kids understand that the responses are invented answers, not human advice. Discuss their boundaries with AI technology. What topics are appropriate to discuss with AI, and which are not?

🗨️ Keep conversations open and ongoing. Remind kids that AI can provide inaccurate information — including about health issues — and should never replace the role of medical professionals. For additional guidance, check out “Teen safety: 6 tips for AI and social media,” an infographic by the News Literacy Project and the American Psychological Association.

➡️ For a baseline understanding of how AI tools affect the information landscape, read this AI infographic.

Related:

“Some Kids Will Never Think AI Is Cool” (Wired)

📖 AI children’s books: While a personalized, AI-generated children’s book might sound like a great gift, the end products are often low-quality slop. Parents also worry about how their kids’ likeness and personal details will be used by AI platforms.

🤖 Bots at bedtime? Meta is developing an app that reads bedtime stories for children, raising questions about the role of tech during parent-child bonding moments.

📵 Social media ban: France recently became the latest country to ban social media for children, following Australia and Malaysia. More countries are also preparing similar bans.

💊 AI health influencers: The New York Times found hundreds of AI-generated wellness “influencer” accounts — some using synthetic personas that appear to be doctors — sharing bogus health claims and selling supplements.

📰 Teens reading newspapers: Researchers studying news consumption habits found that most teenagers will read newspapers when given free access.

Did someone forward you this newsletter?

Which video of artwork is AI?

Three recent viral videos appear to show amazing works of art, but only two of them are real and one is an AI-generated fake. Can you spot it? Click on the video below to test your skills.

Reader poll

Last month, we asked if you and your kid(s) have talked about AI-generated content. Most of you (59%) said you talk about it regularly and some of you (41%) said you’ve talked about it once or twice.

👉 Next question: How often do you talk about current issues and events in the news with your kids?

⚪ Every day.

⚪ A couple times a week.

⚪ Occasionally.

⚪ Never.

We will share poll results in the next Scroll Smarter issue.