I think the idea at hand is not that it would be used everyday, but it would be there when needed. What people do everyday is with real data. They are using functions to load the data. Where this would come in useful would be presentations and tutorials.

If leading a presentation on scientific computing in Python to beginners, which would look better on a bullet in a slide?

• np.build('.2 .7 .1; .3 .5 .2; .1 .1 .9'))

• np.array([[.2, .7, .1], [.3, .5, .2], [.1, .1, .9]])

The default way of defining contrived arrays by passing lists of lists is awkward for beginners. While lists of lists are not a hard concept, it's not something you want to force on someone who doesn't know the Python language yet. The second bullet above doesn't represent the readability of the Python world.

I would suggest that this be named np.build() (or np.helpers.build()) in light of it providing a simple interface to building arrays. Again, when you work with real data you are taking an extra step to think about how you load that data. That's not what you need to think about when being introduced to NumPy.