Does AI Make Kids Less Creative? It Depends on Who Is Making the Decisions

TL;DR

AI does not automatically make kids more or less creative. What matters is which parts of the creative process the child still owns.

If a child asks AI to “make me a story about a dragon” and accepts the first result, most of the interesting creative decisions have been outsourced. If the child invents the dragon, decides what it wants, asks AI for a few possible problems it could face, rejects two, changes the third, and writes their own ending, AI is playing a very different role.

The research is still early. Recent reviews find promising evidence that generative AI can support idea generation, storytelling, creative confidence, and problem-solving. They also identify risks around over-reliance, reduced independent exploration, and children accepting AI-generated ideas too readily. Researchers are not yet in a position to say that AI broadly improves—or harms—children’s creativity. Review of children’s creative AI use

A useful rule for parents is simple:

After the AI responds, does the child have more thinking to do, or less?

The same AI can create very different experiences

“Using AI” is too broad a category to tell us much. These children are all technically using generative AI, but they are practicing very different things.

The first experience is not inherently bad. Generating something funny or impressive can simply be fun. But if most of a child’s “creative” AI use looks like the first row, they are getting less practice at generating ideas, choosing between possibilities, changing direction, and deciding when something feels finished.

Those decisions are a large part of creativity. The polished artifact at the end is only one piece of the process.

What the research actually says

There is real reason to be optimistic about technology as a creative partner.

In an MIT Media Lab study, researchers recruited 51 children ages 5–10 to play a creativity game with a social robot; 48 were included in the analysis. Children paired with a robot that modeled creative behavior generated more ideas, produced a wider variety of ideas, and scored higher on creativity than children paired with a non-creative robot. MIT social robot study

That was a social robot rather than today’s generative AI, but the interaction pattern is interesting. The technology did not simply hand children a polished final artifact. It modeled creative behavior, giving children something they could respond to and build from.

Research specifically on generative AI and children is newer. A 2026 systematic review of 57 empirical studies found AI playing both helpful roles—such as Inspirer, Scaffolder and Co-Creator—and constraining ones, including what the researchers called Deskiller. The authors argue that the effect depends heavily on the role AI occupies in the creative process rather than on the presence of AI itself. Review of 57 empirical studies

Another 2026 review of children’s creative AI use found possible benefits for divergent thinking, narrative creativity, and creative confidence. But it also emphasized that the evidence base is still small, methodologically inconsistent, geographically concentrated, and partly based on work that has not been peer-reviewed. Review of children’s creative AI use

So the evidence does not support either extreme.

AI makes kids more creative is too strong.

AI destroys creativity is too strong.

A better conclusion is that the way AI is designed and used changes what children get to practice.

When AI can make creativity thinner

The clearest risk is instant output.

Give a child a blank sheet of paper and ask them to invent a monster. Before the drawing exists, they have to decide what kind of monster it is, how many eyes it has, what shape its body should be, where it lives, which colors belong together, and what to do when the first attempt looks wrong.

Now replace that process with: “Make me a funny purple monster.”

Three seconds later, a polished monster appears.

The child may love it. That delight is real. But the creative activity has changed. They did not have to wrestle with most of the decisions that produced the result.

This matters because creativity is not only about generating ideas. It also involves selection, evaluation, revision, and judgment. If AI repeatedly makes those decisions before the child gets a chance to, the child simply gets less practice doing them.

The concern is not that AI output is “too good.” The concern is that AI can collapse the distance between wanting something and receiving it so completely that very little is left to figure out.

There is an interesting parallel in adult research. In a randomized experiment on short-story writing, people given AI-generated ideas produced stories that were judged more creative, better written, and more enjoyable. But the AI-assisted stories also became more similar to one another. In other words, AI improved individual outputs while reducing collective variety. Short-story experiment

That study involved adults, not children, so we should not assume the exact same effect applies to kids. But it raises an important question: if everyone starts from the same machine-generated possibilities, do individual creations improve while our ideas slowly converge?

When AI can make creativity easier to access

The other side matters just as much.

Children often have ideas they cannot yet execute. A child may imagine an elaborate story but struggle to write quickly enough. They may know exactly what a character should feel like but have no idea how to draw it. They may hear a melody in their head but lack an instrument or the technical ability to reproduce it.

Sometimes the idea disappears because the gap between imagination and execution is simply too large.

AI can narrow that gap.

It can help a child find words for an idea, suggest possibilities when they are stuck, turn a rough concept into something they can react to, or remove a technical barrier that was stopping them from continuing. Recent reviews suggest that carefully scaffolded AI may help children generate, express, and refine ideas, while also warning that too much automation can weaken agency and independent exploration. Review of AI’s creative roles

The useful distinction is not therefore AI vs no AI. It is what job we give the AI.

There will be times when every one of these modes is fine. The problem appears when Generator quietly becomes the default definition of creativity.

AI should not always go first

Timing matters too.

Imagine asking a child to invent an animal that could live inside a volcano. If AI immediately shows them four beautiful volcano creatures, those examples may shape the child’s imagination before they have generated anything of their own.

A stronger sequence is often to let the child make the first move. They sketch or describe something. AI enters after an idea exists, perhaps suggesting a strange adaptation, asking how the creature eats, or proposing a problem it might encounter. The child then decides whether any of those ideas belong in their creation.

This does not mean examples are bad. Children learn from examples constantly. It simply preserves moments where the child has to create a possibility before the machine provides one.

Sometimes the most useful AI intervention is waiting.

The easiest test for parents: who made the decisions?

Parents do not need to understand model architectures or prompting techniques to judge whether an AI activity is creatively useful.

Ask your child what they made and listen to how they explain it.

“I didn’t like the first idea, so I changed it.” “It suggested three endings and I combined two.” “I told it that the monster should be scared instead of scary.” “The first version didn’t feel right, so I tried again.”

Those explanations reveal a child who is still evaluating, rejecting, choosing, and revising.

If the whole explanation is essentially, “I asked AI to make it,” then the machine probably did most of the creative work.

That is fine sometimes. Not every minute of screen time needs to become developmental training. But if the goal is creative practice, the question is how much of the thinking still belongs to the child?

Where Taroo fits

This is the design problem we care about most at Taroo.

Adult AI products are usually optimized for efficiency. You know what you want, you type a prompt, and the system tries to get you to the finished result as quickly as possible. For an adult trying to finish work, that can be exactly right.

For a child learning to create, the thinking that gets removed may be precisely the thinking they need to practice.

Taroo instead begins with guided creative quests across storytelling, drawing, music, speaking, and observation. A quest gives the child somewhere to begin, but it does not determine what their creation should become. AI can then respond, introduce another possibility, give feedback, or help them iterate.

The distinction is subtle but important. Instead of asking AI to “write me a story,” a child might begin with an unusual picture, an unfinished situation, or a strange character. They decide what is happening first. AI might later ask a question or introduce a twist, and the child decides whether it belongs.

The goal is not to keep AI away from children. It is to stop AI from removing all the interesting decisions before the child has had a chance to make them.

What parents can encourage

You do not need to ban image generators or panic when your child asks an AI to make something ridiculous. Instant generation is fun, and experimenting with these systems is part of learning what they can do.

What matters is variety. Sometimes let the child create before consulting AI. Sometimes ask for three possibilities rather than one answer. Encourage them to reject suggestions, combine ideas, explain why they chose something, or take an AI-generated starting point somewhere the machine did not expect.

Most importantly, let them get stuck.

Being stuck is not evidence that a creative activity has failed. Figuring out what to do next is part of learning how to create.

So, should kids use AI creatively?

There is a downside to simply keeping children away from AI too.

They are growing up in a world where creative AI will increasingly be ordinary. Learning how to direct these systems, recognize generic outputs, question suggestions, develop taste, and retain ownership over an idea may become important creative skills themselves.

AI can also unlock forms of expression that previously required substantial technical skill. A child who cannot animate can explore animation. Someone who cannot yet play an instrument can experiment with musical ideas. A child whose handwriting slows their imagination can explore a story without being constrained by writing speed.

As execution gets easier, however, judgment becomes more important.

When producing something takes seconds, the difficult questions become: Is this actually interesting? Does this feel like mine? Which version is better? What should change? Why?

Those are creative questions too.

The takeaway

AI does not make children less creative simply by being present. What matters is which parts of the creative process we give away.

If AI continually supplies the idea, execution, judgment, revision, and finished result, children have fewer opportunities to practice those things themselves. If AI provides a useful spark, asks a good question, removes a technical barrier, or helps a child examine an idea they already started, it can expand what they are able to make.

So instead of asking “Did my child use AI?”, ask:

“Who made the important decisions?”

The goal is not to keep AI out of children’s creativity.

It is to keep the child inside it.

Frequently asked questions

Does research show that AI makes kids less creative?

Not conclusively. Recent reviews find both potential benefits and risks. AI may support divergent thinking, storytelling, creative confidence, and problem-solving, while over-reliance can reduce independent exploration or creative agency. Researchers consistently describe the evidence as early and context-dependent. Review of children’s creative AI use

Can AI help kids become more creative?

Potentially. Research with children has shown that artificial agents can support creativity when they model creative behavior and scaffold children’s thinking instead of merely producing finished work. MIT social robot study

Should kids use AI image generators?

They can. The stronger creative experience is usually one in which the child contributes ideas and judgment rather than repeatedly requesting finished pictures. One useful approach is having the child develop an idea first, use AI to explore variations, and then decide what to keep, reject, or change.

Is prompting a creative skill?

It can be. Writing a strong prompt requires intention and articulation. But prompting is only one piece of creativity. Children also need opportunities to originate ideas, compare possibilities, revise work, and develop their own judgment.

How can I tell whether AI is doing too much?

Ask the child what decisions they made. If nearly every interesting idea, execution choice, and revision came from AI, little creative work remained for the child. If they rejected, modified, combined, and developed AI suggestions, the tool is playing a more supportive role.