Using agents for real problems

I read the WSJ article over the weekend that's kicked off the "AI killing humanity" coverage. Lots to unpack in the article but one thing I found interesting is how solving math problems are used as marketing. The section on spend stood out to me... For those that don't have WSJ: The wicked Navier-Stokes problem had vexed mathematicians for nearly a century. AI needed millions of dollars in computing resources and a few days. I've read Machines of Loving Grace. I understand the challenges Dario lays out about physical world dependencies/bottlenecks in biology and neuroscience but why wouldn't we still start there? Forget math proofs for now... there's got to be low hanging fruit in infectious disease or understanding the human brain that would serve as more useful marketing and be beneficial to humanity, no? Or, if we're locked into math as the only domain we'll touch, why not focus on Physics and better understand the universe?