There is a reasonable concern behind the question: if AI writes SQL for you every day, will your SQL skills disappear?

The answer depends less on whether you use AI and more on how you use it.

Use AI to explain, not only generate

Ask it to explain joins, window functions, execution logic and alternatives. Then read the query yourself.

Predict the query before asking for it

Even a rough mental outline gives you something to compare against. If the generated SQL is radically different, investigate why.

Debug with hypotheses

Give AI the problem and ask for small tests before requesting a rewritten query. This keeps you involved in the reasoning.

Rewrite important queries yourself

For production logic or business-critical metrics, treat AI output as a draft. You should be able to explain every meaningful part of the final query.

Keep the fundamentals

Grain, joins, filtering, aggregation, NULL behavior and date logic remain analyst skills even when a model can generate the syntax.

This is one of the core ideas behind my first ebook on becoming an AI-assisted analyst.