Your preschooler thinks the talking speaker is a little bit of a person — and that's completely normal. Here's what young kids actually understand about AI that talks, and the simple things to say back.

Sooner or later a young child asks it out loud: "Is Alexa alive?" It's one of the most honest questions in parenting, because to a three- or four-year-old the answer isn't obvious at all. A thing in the kitchen hears you, talks back, knows answers, plays your song, and has a name. Of course it seems a little alive.

(We'll use "Alexa" as shorthand here, but the same idea applies to Siri, Google Assistant, Gemini, other smart speakers, and kid-facing chatbots — anything that talks back.)

For this age, the first AI lesson isn't how any of it works — that's years away. It's something simpler and more important, and it's worth pinning to the fridge:

For kids ages 3 to 6, AI usually shows up as a voice in the room. This guide covers what young children really think a talking device is — and the simple words that help you explain it.

It's not a mistake to shut down — it's a normal stage to gently guide. Researchers who study early childhood and technology have found that voice assistants appeal to young children as both information sources and social partners — the very fact that it talks like a person makes a small child relate to it like one. That social pull is exactly what makes a talking device more confusing for a preschooler than a screen.

And children genuinely overestimate what these devices are. In a study of 166 children ages 6 to 11, many overestimated the smart speakers' intelligence and were uncertain about whether the devices had feelings or agency. If six- to eleven-year-olds are still working it out, a three-year-old is starting from "it's basically a person." Your job isn't to make them feel silly for thinking so — it's to gently, repeatedly draw the line.

Guidance for the birth-to-eight years, developed by early-childhood researchers, points the same direction: at this age AI use should be adult-mediated and grounded in real, relationship-driven human communities — not a child alone with a device. The same guidance notes that many AI tools were not designed for children, yet children use them daily, and that young kids simply can't be expected to understand how AI works. That's the case for co-use in one sentence: they can't interpret it alone, so you interpret it with them.

There's early, limited evidence pointing the same way. A small preprint study with children ages 5 to 6 suggested that parent co-presence may help children make sense of their interactions with AI. It's preliminary work with a small sample — not proof — but it lines up with the broader developmental advice: the most powerful safety feature for a preschooler isn't a filter, it's you in the room.

You don't need to explain how AI works — you just need a few warm, simple lines you can repeat. Google's own guidance for parents makes the key point: kids should understand that AI is not a friend. Here's what that sounds like in everyday words.

When they ask if it's alive: "It can talk, but it's not alive like you and me. It's a talking computer — it answers questions, but it doesn't have feelings, and it isn't your friend, even though it's fun."

When it gets something wrong: "See, it made a mistake! Computers can be wrong. That's why we check with a grown-up."

When they want to tell it a secret or talk to it like a buddy: "That's a talking computer, not a person. Private things are for trusted grown-ups, not devices."

Do: use it together · call it "the talking computer" · point out mistakes calmly · keep private things for people. Don't: call it your child's friend · let it become a private companion · treat its answers as always right · let young kids use it alone.

A few signs that a young child is over-attaching or misunderstanding the device — none are alarming on their own, but each is a good cue to gently re-draw the line and stay closer during use: they say the device is sad, lonely, angry, or has feelings; they apologize to it, or worry about hurting its feelings; they want to talk to it alone, or prefer it to people; they believe it over you when the two of you disagree; they tell it private family information.

The line you draw at 3 to 6 doesn't disappear; it matures. As kids move past this age, "it's a talking computer, not a person" grows into the broader family AI safety talk — that AI is a tool not a person, that some things don't go in it, that it can be confidently wrong, and that big feelings go to a real person. Getting the first version right at three makes the fuller version at eleven feel obvious rather than new.

📎 Free download: "Is Alexa Alive?" — a one-page guide with the one thing to teach, the warm scripts, do/don't, three habits, and the gentle flags to watch for.

"Is Alexa alive?" isn't a problem to shut down — it's the first real AI conversation you get to have, and it's a lovely one. The answer is warm and simple: it's a talking computer, it's not alive, it doesn't have feelings, it can be wrong, and we use it together. Say it kindly, say it often, and stay in the room. At this age, your presence is the whole curriculum.

This article is general guidance and is not legal, medical, or mental-health advice. Disclosure: AI by Age is run by people who work in AI and build AI-related products. We take no AI-vendor sponsorships or affiliate payments. Full disclosure on our About page.