What to look for in AI adaptive learning tools that change for your child

A real adaptive learning tool changes something specific because of something your child did, and can tell you what and why. Look for four things: it records what your child does, it has a clear rule for what changes, it changes the right things and not the safety rules, and it shows you in plain words what it saw. Peabody (peabody.kids), a reading and learning app that is free through the end of 2026, works this way for its words and stories, and we use it as the example below. The checklist works on any tool.

"Adaptive" and "personalized" are used for everything now, from a tool that remembers your child's name to one that rebuilds a lesson every minute. The word alone tells you nothing. This guide shows you how to look underneath it.

What "adaptive" should mean

Strip away the marketing and an adaptive tool does this loop:

1. Observe what your child does (what they got right, wrong, skipped, how long it took).
2. Decide using a rule, based on those observations.
3. Change something: the next item, the level, how soon a word returns.
4. Show the grown-up what happened.

If any of the four is missing, it isn't really adaptive. A tool that observes but never changes anything is a tracker. One that changes things without observing is a random shuffle. One that changes things but never shows you why is a black box.

Part 1: What should be able to change

Good candidates for change, because they depend on the child:

• Level. Easier or harder material, in small steps.
• Which items come back. Words or facts a child missed should return sooner than ones they know.
• Pace. How much is offered at once, and when to stop.
• Interests. The topic or characters, chosen by the child.
• Support. Hearing a word, a hint, reading together.

For example, in Peabody a word only counts as sticking after your child reads it right on three different days in a row. A miss puts it back to practising, and missed words come first. When the last few books have all been too easy, or all too hard, it offers a five-tap check at the level next door. Those are three concrete changes with three readable rules.

Part 2: What should not change

This part gets skipped, and it matters most for children. Some things should be the same for every child, whatever the AI "learns":

• Safety rules. What a story or answer may contain doesn't relax because a child asked.
• Privacy. What is collected shouldn't grow as the tool "gets to know" your child.
• Honesty. The tool shouldn't flatter. It shouldn't call a miss a win or invent praise.
• No comparison. Adapting to your child is not ranking your child. Be wary of leaderboards and percentiles aimed at young children.
• The end. A session should be able to finish.

If a tool says it adapts to keep a child engaged for longer, ask whether the aim is learning or time on screen.

Part 3: A five-question test you can run in an evening

You don't need technical knowledge. Use the tool with your child for two or three short sessions, then ask:

1. Did it change after a mistake? Miss something on purpose, or let your child miss it. Does it come back later, soon, in a different form?
2. Did it change after a run of easy ones? If everything is right, does the level or pace move up, or does it keep serving the same thing?
3. Can you find the rule? Look in the help pages, the grown-up view or the blog for a plain statement of what changes and when. "Three right days in a row" is a rule. "Our proprietary engine" is not.
4. Does it say what it can't do? Honest tools state limits: no diagnosis, no hearing your child read, no grade comparison.
5. Does the grown-up view match what you saw? If it says a word is "sticking" and you saw your child miss it twice yesterday, ask why.

Part 4: Where AI belongs, and where code is better

Two different jobs get called "AI" and they need different tools.

Peabody uses AI to write stories, and code to measure them, set levels and track words. A tool that uses a chat model to judge your child's level or decide when a skill is done is making a more fragile choice. You can read more in how AI-driven adaptive reading works.

Part 5: Claims to be careful of

A small tool that says "we can only see taps, so this is what we can and can't tell you" has earned more trust than one that promises everything. Where is my child stuck? shows what honest reporting looks like.

Part 6: What to ask the school or teacher

Adaptive tools are most useful when they match what your child is doing at school. Ask the teacher:

• What level or range of books is right for my child right now?
• Which words, facts or skills should we practise?
• Is there anything you'd rather we didn't use?

Then try the tool a step either side of that level and see what feels right. Use your child's teacher as the check on any tool's view.

A one-page checklist

• [ ] It tells me what it observes
• [ ] It tells me what changes, with a rule I can read
• [ ] It changes level, review or pace, not safety or privacy
• [ ] It returns missed items sooner
• [ ] It moves on when things are easy
• [ ] It shows me, in plain words, what my child did
• [ ] It states what it can't do
• [ ] It doesn't rank my child against others
• [ ] A session can end
• [ ] My child asked to use it again, and I'm fine with what they saw

If you tick eight or more, you have a tool worth keeping. For starting points, see adaptive learning for beginning readers, or try a free Peabody story and run the test above yourself.

Learn more elsewhere

• Stanford SCALE: the evidence base on AI in K-12: a 2026 review: of more than 800 papers, about 20 were strong causal studies, and hint-giving tutors did better than answer-giving chatbots.
• OECD Digital Education Outlook 2026: doing a task with AI isn't learning it; gains often vanish when the AI is taken away.
• Common Sense AI risk assessments: independent tests of the chatbots and AI products kids actually use.

For about seventy more, checked and sorted by who they're for, see where to learn about AI.

Questions parents ask

What is adaptive learning for kids?

It means the material and practice change based on what a child does: level, which items return, and pace. A good tool can tell you which of these changed and why.

How can I tell if an app is really adaptive?

Let your child miss something, then see if it comes back. Let them get a run right, then see if it gets harder. Then look for a plain statement of the rule.

Does adaptive learning replace a teacher?

No. It adds practice. It can't hear a child read, diagnose a learning difference, or know what a classroom is doing.

Is AI needed for adaptive learning?

No. Adaptive rules like spaced review have existed for decades without AI. AI is most useful for writing fresh material. Many of the best adaptive decisions are plain, readable rules.

Is Peabody adaptive?

Yes, in specific ways: stories are written at the level a child picks, practice words return by a three-right-days rule, and the level is checked when recent books were all too easy or all too hard. It doesn't rank children, and it's free through the end of 2026.