One of the biggest problems in mathematical research is that it takes time, and for me, an enormous amount of time is wasted searching for results, and by results, I mostly mean references but not the proofs.
Here is a hypothetical situation. Suppose I am learning the relationship between curvature and topology. I know what curvature is, but I don't know how it relates to topological properties. For example, does positive curvature leads to compactness? Or does they are unrelated? The obvious thing that one can do is to find books. However, in this particular example, some of it can be found in the book A Panoramic View of Riemannian Geometry, which is 1600 pages or so. However, this book has the advantage that it has omitted proofs. Even so, reading this book is not easy since different books use different notations and assume different preliminaries.
EDIT: As I expected, many would argue the importance of proofs. However, here is a real situation. I am looking at AI-assisted drug design in something called geometric deep learning combined with diffusion process. In the field of AI, you don't have one or two years to prove something; instead, you have months or even weeks. I know many people look down on AI research or even call it pseudo-science or paper factories, but drug design actually does save money, and the beauty is that this field applies quite deep mathematical results to real-world problems. In this humble person's opinion, there is no shame in that.
On the other hand, I do believe you have to do some proof in order not to lose the ability.
These three are very important.