So, in my grad program in ag, we had a Bay-area ML startup come in to give a seminar on how they were revolutionizing agriculture. The presented their findings on how to increase yields (which they claimed could only be understood from their algorithm).
The problem they were diving into was well understood, and has been researched to death for the last 100+ years. And they had the relationship backwards, not understanding their "input" to increase yields was actually a response to low yields. They were the opposite of helpful, but rather a waste of our time.
As with anything, it helps to know the current state of knowledge before you jump into contribute. An understanding of math doesn't get you there.
The problem they were diving into was well understood, and has been researched to death for the last 100+ years. And they had the relationship backwards, not understanding their "input" to increase yields was actually a response to low yields. They were the opposite of helpful, but rather a waste of our time.
As with anything, it helps to know the current state of knowledge before you jump into contribute. An understanding of math doesn't get you there.