Python is used heavily in data science (and a lot of other places) because people who go to university for non software engineering disciplines get taught Python because it's "the easy language that already has libraries for this research we're doing." Those people then go on to write more of these libraries. Their code does amazing things, but very slowly.
There's a good video series called "Programming Paradigms" by Jerry Cain, taken from his class at Stanford. I'm not sure how long ago it was, but it was before whiteboards, when they were still using chalk. He just started including Python that year when it was the up-and-coming thing, as an example of a higher-level language that does a lot of stuff for you. It probably seemed like a breeze for the students after the previous weeks spent on C, assembly, and Lisp, but at least they got some of the fundamentals of how things worked first.
Totally, it's always about trade-offs. It takes a decent amount of time programming to become comfortable choosing the language based on the task rather than tailoring the task to the language.
