Yes, when people in the 4000s write clickbait holonet articles about how the Master Theorem of Neural Network Scaling was not actually invented by the Muskovites of Mars, but was found written on an Ancient American tablet containing the works of Hoffmann et al., I hope somebody will offer substantial corrections.

That's not in any sense a value judgment of the empirical work done at DeepMind. Nor does it dtop anybody from writing a better article which explains tha the Babylonians (and many others) used the empirical observations underlying the theorem, while explaining that this did not constitute mathematical proof.

Found on an ancient tablet as the work of Schmidhuber et al. but ok

Pointing out that it's not a theorem doesn't mean it's been trivialized.

If there's no proof, it's not a theorem, by definition. That doesn't make empirical observations worthless -- far from it!