There are a number of flaws with your argument:
1) "Black kids" most certainly are "less able to program," at least in the US, as revealed by the empirical evidence.
2) If you meant to write that they were congenitally not "less able to program" (which is how I interpret it), then it is a supposition based on speculation and not upon empirical evidence. The fact is, we do not know to what extent congenital factors affect ability in computer science nor do we know if they are unevenly distributed along gender, ethnic, or racial lines.
3) African Americans and "blacks" are two different groups.
4) If you had "race-blind" programs than there would be no way to target the demographics most underserved. The hill-folk in rural West Virgina and the impoverished people in Bedford–Stuyvesant both tend to be poor and undereducated and are at higher than average risk to be the victim of a crime, but for a police/sherrif's department to develop the same strategies to combat the higher crime rate in those very different demographics would be laughably obtuse.
Likewise, if you're trying to get poor, mostly rural white people in the Ozarks into computer science, you need a very different strategy than you would to get poor, mostly Latin kids in San Ysidro. Ignoring essential demographic information would be tantamount to incompetence.
Also, if helping one race to the exclusion of other races is "racist", then our whole society and culture is racist, as there exist many social institutions, formal or informal, that create that effect. It seems kind of silly to worry about Google giving money to programs that help low-participating demographics achieve parity when there exists a massive institution called American society that exists to elevate members of one population above another, on gender, racial, ethnic, national, and pecuniary lines.