← Back to recent questions

Any of the big chatbots will give a good answer to "why is the sky blue". The more interesting question is whether you want them to. There is something a bit sad about a child asking the family device, and the device answering, while the parents stay in the kitchen. The whole point of those questions, when I was small, was that they got me a parent's attention for a few minutes. Worth a moment's thought about how to use this with your kids rather than instead of you.

How I'd approach it

Do it together. The setup that works for me is the phone in a holder on the bench, ChatGPT open in voice (chat) mode, and I am in the conversation as much as the child is. The AI answers, then I add the bit it missed, or the bit that is actually relevant to where we are or what we just did, or the question my child should ask next. It is a third voice in a conversation between the two of us, not the voice of authority that ends the question.

For the voice mode itself, both ChatGPT and Gemini are very good in 2026, on the free tier. Pick the one that matches the family phone. If it is an Android with Gmail on it, Gemini is already there. If it is an iPhone, install the ChatGPT app. Its voice mode is the most natural-sounding of the lot. There is no need to install three different apps and "compare" them for this kind of use.

Two settings worth changing once you have picked one. In the app settings, find the option about training on your inputs and turn it off. You probably do not want anything your child says to a chatbot used to train a future model. Then, ask the AI itself, in plain language, to "answer my child's questions in short sentences with no scary content and no advertising". It will keep that instruction for the rest of the conversation.

If you want a slightly different flavour for an older child, try Perplexity. It is free, and instead of just answering, it shows the websites it pulled the answer from. For a curious nine or ten-year-old who is starting to ask "but how do you know that?", that is genuinely useful. It teaches the habit of looking at where an answer came from rather than just trusting the voice on the phone.

What I'd avoid

Two things to flag. The first is the catch with voice mode. Chatting in voice burns through the daily free-tier quota faster than you would expect. A long conversation with a curious child can hit the limit in fifteen or twenty minutes, especially if the child gets going. If that becomes a regular pattern, either keep sessions shorter, or decide whether the paid tier is worth it for you.

The second is treating the AI as the last word. Chatbots still make confident mistakes, and a child will believe whatever comes out of the speaker the first time. For science basics like the sky, the food chain, or the planets, the answer is almost always fine. But if your child asks something where being wrong matters (a medical question, anything about safety, anything about a real person), the same rule applies as it does for the rest of us: check the answer before passing it on. Most of these tools officially require a user to be 13 or older, and the easiest way to honour that is to treat the chatbot as your account that the family uses together, not a private tool you hand the child.