Standardized tests can include unintended biases. Anytime there is a cultural difference, there will be vocabulary differences including subtle meaning differences that can skew responses in standardized testing. I haven't looked into the details, but it makes sense.
I know there are studies that show what the correlation levels are between college performance and standardized test scores, but I don't know if that data is broken down by different groups; I wouldn't be surprised if the correlation is the strongest among well-to-do white students and the weakest among poor students of color.
Should an admissions office use all the available data to generate the best possible calculation on a potential student's probability of success? That might include standardized scores, grades, data on the school they come from, their economic conditions, and their race and ethnicity.
And then there's the question of whether they want to be completely fair. There's a good argument that minorities are disadvantaged by generations of discrimination, so efforts to reverse that are still needed. Whether college admissions is an appropriate place to work towards that goal or not is a matter of debate.
I'm not sure what the right answers are. Having several years of not using standardized test scores should help give more data on what does and doesn't work in admissions.