Why, in the age of AI, should CS students should invest in learning math?

AI recently won a gold medal in the international math olympiad, a really hard math contest.

Given that, why should you bother learning math? Why not just use AI to "solve math" whenever you need it? Let's take this theme in a series of questions:

Why should CS students learn math when AI can do math?.

It is coding skill that is most at risk, rather than math.
Also, let's define "math" carefully. Yes, "doing" some math (derivations, solving already-framed problems) can be done by AI. But what about "thinking mathematically"?
The skill that will be important in the future is being able to think abstractly. And so, it's the skill of thinking mathematically that will matter.

Won't AI be able to think mathematically as well?. Generative AI is able to extend slightly what is already fed into it, through a process of randomization. However, businesses and the scientific enterprise will need people to understand the results of AI's mathematical exploration, to formulate new abstractions, and to understand how to apply it to the real world. This is likely to come from humans who are mathematically trained. For example, going to back to when cryptography was first devised (by humans), would AI have devised the core ideas of cryptography all by itself when none of it was present in the "math that AI trained on?"

Fine, some mathematically trained people will be needed. Why should that be you?. History has shown that as lower-level skill has been commodified and made more mundane or automatable, the value-add for people is at higher levels of abstraction. Application areas that need mathematical abstraction has always been, and will continue to be, important. In fact, AI itself is based on various areas of math. That is, working in an AI team requires familiarity with some mathematical areas.

But what if I just need to check off a course requirement? One can understand the mindset of "I just need to get past this course because it's a requirement". There is in fact a modest-effort pathway to passing the course with this mindset. But consider adopting a growth mindset and an opportunism mindset. That is, if there's any time in one's life to adapt and grow, it's during youth. What exactly is wrong or what will be lost with going "all in" and trying that out?

Other reasons why investing in math has long term payoff:

The core areas (discrete math, linear algebra, probability, calculus) have proven to have long-lasting value.
- Consider the story of AI itself. For the longest time, in the 1990's and early 2000's linear algebra and probability were generally "it's good for you" vitamin-like courses. But AI, data science and quantum computing changed all that.
- Both linear algebra and probability are essential if you want to work in AI development or data science. And linear algebra is at the very foundation of quantum computing.
Over the past few decades, various areas of math have risen or waned in their "CS importance" but all have proven extremely useful over the long term.
Mathematical training sharpens the intellect. Just like cardio training is good for any sport, math training develops a general comfort with abstraction and being able to handle difficult problems.
Software developers in industry say that it's important. In a recent survey of industry professionals, most of whom work in software development, 90% think that mathematics should be required for CS students, especially the core areas of discrete math, calculus, linear algebra, and probability.
It's never too late to start investing. Students often mistakenly think "Math is not for me" or "I don't have the math gene". This is as absurd as saying "[Fill in foreign language] is not for me" or "I don't have the [Fill in foreign language] gene". Yes, it's challenging and yes some other got a head start in elementary school (like those who had [Fill in foreign language]). But so what? If you're not 70 years old, it's not too late, even if it means stumbling a bit early on.
After a certain point, it's easy to feel progress. There are many skills where, even after a long investment, it's hard to feel you're progressing. Math is not among them. Initially, there's a bit of pain in getting up to speed, but after that, you will feel noticeable progress. It's important not to confuse this progress with grades: they are two different things. It's quite possible to make considerable progress yet not have the grades to show for it.
Math is aesthetically beautiful. After math starts to become comfortable and second nature (which does take several courses), later math opens a world of beauty. There's much pleasure to be had in exploring mathematical ideas and observing their power in exploring the world. As a computer science example, consider Turing's landmark result: some problems are simply unsolvable by computer. This is a deep mathematical result that, inspite of its signifance and proper place in the math pantheon, has a proof that is quite approachable in an undergraduate course.

So, then, how to "invest"? What does it mean?

Rule #1: no shortcuts. It's a skill, which means doing it yourself. No AI, no unnecessary help. The struggle is necessary in developing skill.
Rule #2: be patient. Any abstract skill development takes time. It can't fully happen in a single course. It's a long term commitment that itself will have some ups and downs along the way.
Rule #3: take many math courses. Just like learning a new programming language helps you become a better developer, every math course adds to your skill and mathematical development.
Rule #4: repetition is king. It's just how our brains work. If you don't understand something, that's perfectly fine. Just try again and again.
Rule #5: get guidance not assistance. Guidance from mentors and professors but not assistance from you-know-what.