I am for the version on the right. Here's why, from a Typographic, Design Element, and Human perspective.
From a Typographic Alignment Perspective
In the examples I show, I have stripped the input box so you can see it purely from a typographic perspective.
Visual hierarchy looks like this:
See me first
- See me second
See me third
- See me fourth
Notice how the information is staggered.
If the user wanted to quickly sift through all the labels it would look like this:
See me first
- Ignore me
See me second
- Ignore me
Having them visually separated by space, makes it easier for the person to visually parce what's what. If all the text is aligned, it will get read like this:
See me first
See me Second
See me third
See me fourth
All the text blurs together and there's no easy way for the user to skim the content.
From a Design Element Perspective
List item
The other way to look at this is by shape. If you abstract it, any words in the boxes are part of the shape of the box. Therefore, aligning text with the input box makes the most sense.
Here's an example of alignment of copy with text in a box. It's usually aligned by shape.

I've never seen any case where the text is aligned with the text in the button.
I understand that in the example it's NOT an input box but a button but in principle it's by shape.
Here are examples:

Google made this leap too. Here you can clearly see their use of the Grid to clean up the design.

From a Human Perspective
When filling in forms, we naturally leave a space before writing.

Source: My experience formally trained as a Graphic Designer and having worked in the field for many years.