It was to highlight that Math could also be introduced by ways of pattern recognition. And that it may be an appropriate time to delve into other such methods, and look beyond rote learning.
Not everything is pattern recognition. To be honest, hearing this repeatedly said is flashing me back to Umi Zumi with their box-ticking educational requirement...
Taking the method you mentioned, IF a student could demonstrate a "simpler way" by use of another method when certain parameters are concerned, would they be in good steed? I guess they would need to do this in the classroom, not exam time?
For the vast majority, they need to be able to demonstrate that they understand the method being taught. The point of teaching a variety of methods is to give children the tools they need for the future. If they only know one way (eg. Completing the Square or only using the formula) then they don't have all the tools.
This applies in so many situations - I'm a programmer, and perhaps someone I work with can sort a list using quicksort, and it'll work, but if they don't know any other way, they don't know if that's appropriate for the situation, they might (likely are) missing much better (faster, more efficient) ways of doing something.
In teaching, including subjects with correct answers, the point of the exercise isn't just to get the answer, but to know how you got the answer, so you can apply it to other situations.
And as to AI - yeah, I've worked with it - it's useful for some things, it's good at some things, but it won't be replacing the people who have put in the effort (and have the ability) to get an answer multiple ways, to evaluate the options and pick the best way for the situation, and anyone relying on it to get their answers for them is setting themselves up for a massive fall.