[Jugurtha]

I think fast.ai[^1] would be a good start. I think you'll have analysis in high school and learn a bit on that. Khan Academy[^2] seems to have courses catering to a broad spectrum of skill levels in mathematics.

[^1]: https://www.fast.ai

[^2]: https://www.khanacademy.org

[s1t5]

I would advise against fast.ai - "import * from my_custom_library and then watch me run through all the cells in this notebook" is not a good way to learn for anyone at any level.

+1 for Khan Academy if OP hasn't done any linear algebra or calculus.

[Jugurtha]

I understand. The recommendation is tailored to lennyscales who is 16 and wants to get started with ML without linear algebra or calculus in their toolbox yet.

This is interest: a small, delicate, flame. They already got started watching videos, and want to ramp it up with more details. I think the increase in complexity ought to be gradual and match their toolbox: not too trivial to cause boredom, and not too complex to completely discourage them. I think actually getting started, not quitting, and picking up the necessary tools along the way to solve problems as they arise is one approach.

They will equip their toolbox with calculus, linear algebra, statistics, and all the things that brilliant people spent lifetimes building on top of each others' work throughout humanity's existence. They may even fall in love with these disciplines and pursue them for their own sake, but it is not lennyscales' ask. Mastering these is not necessary to get started with ML, no more than mastering computation theory and symbolic logic are necessary to get started with programming.

This reply is given with bold assumptions on what you meant in yours, since it was an objection against an approach without a recommendation for another. My assumptions could be wrong.

[lennyscales]

so how would you approach learning ai in terms of coding, not the math behind it?