What an AI actually knows about your child (nothing, and why that matters)
An AI chatbot does not know your child's name, school or reading level unless you type it in - and treating every chat as a stranger meeting for the first time is the safer habit to build early.
A father at one of our Saturday demos opened a brand-new chat and typed "what does my son find difficult in maths" before we had explained anything about how the tool works. He wanted to see if it would say something clever. It said, correctly and immediately, that it had no information about his son at all. He looked almost offended, as if the machine had failed a test it was never sitting.
That answer is exactly right, and it is the correct place to start everything else in this pathway. An AI like the one on your phone does not know your child's name, their school, their reading level or their fear of long division. None of that exists for it unless somebody types it into the box in front of it, in this conversation, today.
What it actually knows
It has read an enormous amount of general text, so it knows a great deal about children in general - what a typical ten-year-old finds confusing about fractions, which books get assigned in which grade, how a school year is usually structured. That is real knowledge, and it is genuinely useful. It is just not knowledge about your child. It is knowledgeable about children and ignorant about yours, which is a strange, lopsided kind of intelligence for a parent to get used to.
Does it remember me from before?
Some assistants now offer an opt-in memory feature that saves small facts across separate chats, and if a family has turned that on deliberately, some carry-over is real. But that is a specific, visible setting a parent switches on - not something every AI does quietly in the background by default. The safer working assumption for a child, especially with a tool your family has not configured, is that nothing carries over unless you can point to the setting that says it does.
Why "it knows nothing" is the safer assumption
Children who assume an AI already understands them tend to overshare without noticing - their school's name, their address, a password they are proud of remembering, a worry they would not say out loud to a stranger on a bus. And a chat window is, functionally, a stranger on a bus who happens to type very well.
Where the gap gets filled on purpose
None of this is a flaw to work around. It is the exact seam that level 3 and level 4 of this pathway are built to fill - deliberately, visibly, and under your child's control, rather than an AI quietly assuming a familiarity it has not earned.
The next article in this level covers what happens the moment you close the tab, which turns out to be an even bigger gap than most parents expect. After that, level 3 shows exactly what to write down so an AI stops needing to be told the same five facts every single time.
Questions we get asked
Does ChatGPT remember me between conversations?
By default, no - each new conversation starts with no knowledge of previous ones. Some assistants offer an opt-in memory feature a family can turn on deliberately, which saves specific facts across chats, but that is a visible setting rather than something that happens automatically and invisibly.
What does an AI actually know about my child?
General knowledge about children their age - typical struggles, common curriculum topics, the kind of books and games kids that age usually know - but nothing personal about your specific child unless it is typed into the current conversation. It does not know their name, school or grades unless you tell it.
Is it safe for a child to tell an AI personal information?
Treat every new AI conversation the way you would treat a conversation with a stranger, because functionally that is closer to the truth than most children assume. Names, addresses, school names and passwords should not be typed in, and a family's own rules about what is and is not shareable are worth writing down - which is the subject of level 9 of this pathway.