I think all these suggestions could help. But it seems highly unlikely that a single company (even a powerful one like Google) can categorically solve the problem through these methods.
The reason I asked about your view of feasibility here is that you seem to be arguing that Google (and others with diversity preference in hiring) are making the wrong choice. But the alternatives you seem to favor are not a real choice Google could execute. It's not like they thought about massively altering the perception and delivery of STEM education to all kids, figured out an effective affordable method to do that, and then just did diversity hiring instead because the VP of HR felt like it.
You could certainly make the argument that there's a better choice they could make, or that their apparent choices are doing more harm than good. But right now you're comparing against an impossible straw-man so it's hard to take the argument seriously.