Should my kid learn coding or AI first?

Neither, in that order — start with a project that uses both, because for a school-age child they are one activity. Building something by directing an AI teaches decomposition, precision, testing and judgement in the same session, and those are the transferable parts of coding. The formal language can come later, when the child hits a limit and wants more control.

Will learning to code be useless by the time my child grows up?

The syntax-typing part is already mostly automated. The judgement part — deciding what to build, specifying it precisely, and knowing when the output is wrong — is more valuable now than it was, because there is far more machine-generated code that needs checking. Whether coding is "useless" depends entirely on which of those two things a class is actually teaching.

Is a coding class still worth the money?

If it is project-based and the child leaves with work they chose and can explain, yes. If it is syntax drills toward a uniform end-of-term output, you are paying for the part machines now do. Ask to see three past students' final projects: if they look identical, that tells you what is being taught.

My child only wants to play games, not make them. Does this still apply?

Often the gap is smaller than it looks. Kids who love games usually have strong opinions about how a game should work, and that opinion is a project specification. Starting from "what would you change about this game" converts consumption into a build more reliably than starting from "let's learn to code."

Is using AI to build cheating at learning to code?

It depends on whether the child is required to verify and explain. A child who reads the generated code, finds the bug, and can say why the fix works has learned more than one who typed the same code from a worksheet. A child who accepts whatever appears has learned nothing — that is the real distinction, not the presence of the AI.

Parent decision guide

Should my kid learn coding or AI?

It feels like a fork in the road: enrol your child in coding now, or wait and let them learn AI? Here's the honest answer a coding school can't give you, because their business is the choice — you don't choose. For a school-age child, coding and AI are the same activity now. The child who learns to build by directing an AI is learning both at once.

Why it's not actually a versus

Modern coding is increasingly "tell a computer what you want, then check it did it." That is exactly what using AI well requires. So a kid who builds a game by directing an AI is practising coding thinking and AI literacy in the same session — problem-solving, precise instructions, testing, and judgement about when the machine is wrong.

The either/or framing comes from a world where coding meant typing syntax from memory. That part is now the machine's job. What's left — deciding what to make and steering the tool to make it — is the durable, teachable skill, and it doesn't split into two subjects.

The exact questions parents ask

Coding vs AI — which first? — Why the two are one skill for a school-age kid, and the order that actually makes sense.
Will AI replace coding? — AI changes what coding is, not whether your kid should learn it.
Is coding still worth learning? — Yes — but the skill shifted from syntax to thinking and directing.
What age should kids start AI? — A practical, age-by-age answer for 5–12s.

Where the question actually comes from

Almost every parent who searches this has had the same week. A coding camp emailed about summer enrolment. A news article said AI will write all the software. A friend's child is doing something impressive with Python. And underneath it sits a real worry that has nothing to do with curriculum: am I about to spend money and my child's Saturday mornings on a skill that expires?

That worry deserves a straight answer rather than reassurance. It is true that the specific thing most coding classes teach — producing correct syntax from memory — is being automated fast, and a class built entirely around that is teaching something with a shrinking half-life. It is also true that the demand for people who can specify what software should do and verify that it does it has gone up, not down. Both things are happening at once, which is why the question feels so hard to resolve from headlines.

The resolution is that "coding" and "AI" were never two subjects. They are two names for the same underlying activity: getting a machine to do something precise on your behalf, then checking that it did. The tool changed. The skill did not.

What the two paths actually teach

Set side by side, the overlap is the striking part. The skills in the middle column are the ones that transfer to anything — and they show up on both routes.

Skill	Traditional coding class	Building by directing AI
Breaking a goal into steps	Yes — core of the subject	Yes — a vague prompt produces a vague result, immediately and visibly
Being precise with language	Yes, via syntax	Yes, via plain language, which transfers more widely
Debugging from a symptom	Yes — the most valuable part	Yes, and there is more to check, because the machine writes more
Judging whether output is right	Rarely taught explicitly	Unavoidable — the AI is confidently wrong often enough to make this a daily habit
Reading code someone else wrote	Sometimes, late	Constantly, from day one
Typing exact syntax from memory	Heavily emphasised	Barely — and this is the part machines now do better

A sequence that works for a school-age child

1. Start with a build, not a lesson The child picks something they want to exist and gets it working, however roughly. This establishes that they are the one with the idea and the machine is the one taking instructions — a framing that matters more than any single concept.
2. Make checking the AI a rule, not a virtue Every session should include at least one moment where the AI got something wrong and the child found it. If that never happens, the projects are too easy or the child is not really testing.
3. Ask them to explain it to you Explaining is where the understanding either exists or is revealed not to. It also converts a screen activity into a family one, which is the difference between AI use you worry about and AI use you can see.
4. Let the underlying code arrive when they ask for it At some point a child hits a limit — they want control the plain-language route will not give them. That is when Scratch, Python or JavaScript becomes a tool they want rather than homework they were assigned.

What the "versus" framing costs you

Treating this as a fork has two practical costs. The first is delay: parents who cannot decide often do nothing for a year, and a year is a long time in a child's interest cycle. The second is that the framing pushes you toward whichever option has the better marketing, which is generally the paid class, not the better fit.

A more useful test than "coding or AI" is: does this activity end with my child holding something they made and can explain? A coding class that ends that way is good. An AI tool that ends that way is good. A coding class that ends with twenty identical projects, or an AI tool where the child watched a machine work, both fail the same test for the same reason.

Quick answers

Should my kid learn coding or AI first?: Neither, in that order — start with a project that uses both, because for a school-age child they are one activity. Building something by directing an AI teaches decomposition, precision, testing and judgement in the same session, and those are the transferable parts of coding. The formal language can come later, when the child hits a limit and wants more control.
Will learning to code be useless by the time my child grows up?: The syntax-typing part is already mostly automated. The judgement part — deciding what to build, specifying it precisely, and knowing when the output is wrong — is more valuable now than it was, because there is far more machine-generated code that needs checking. Whether coding is "useless" depends entirely on which of those two things a class is actually teaching.
Is a coding class still worth the money?: If it is project-based and the child leaves with work they chose and can explain, yes. If it is syntax drills toward a uniform end-of-term output, you are paying for the part machines now do. Ask to see three past students' final projects: if they look identical, that tells you what is being taught.
My child only wants to play games, not make them. Does this still apply?: Often the gap is smaller than it looks. Kids who love games usually have strong opinions about how a game should work, and that opinion is a project specification. Starting from "what would you change about this game" converts consumption into a build more reliably than starting from "let's learn to code."
Is using AI to build cheating at learning to code?: It depends on whether the child is required to verify and explain. A child who reads the generated code, finds the bug, and can say why the fix works has learned more than one who typed the same code from a worksheet. A child who accepts whatever appears has learned nothing — that is the real distinction, not the presence of the AI.

Stop choosing — let them do both

Luchi teaches kids to build real projects by directing Luchi: coding thinking and AI skill in one guided path.

Get on the list

See coding for kids