.
The formula for academic index is public knowledge. It is essentially 1/3 converted GPA, 1/3 SAT/ACT, and 1/3 SAT subject test. However, AI is used by The Ivy League athletic conference to set minimum stat standards for athletes. The basic idea is athletes must average no more than 1 standard deviation below the rest of the class average. Individual key contributor athletes may be more than 1 SD below, but the average for athletes can be no more than 1 SD below the average for non-athletes. As such this minimum AI stat requirement for athletes requires calculating AI for all students.
Non-athletes go through a significantly different process that I expect does not emphasize AI. Harvard (and likely others) do rate applicants on a 1-5 type scale on a number of categories including academics. And the academic rating is well correlated with AI, just as the academic rating is correlated with the individual components of the AI calculation (grades and scores) .
I believe thibault is more talking about stats that are highly likely to result in rejection, rather than automatically being admitted based on just stats. For example, during the lawsuit period, the majority of non-ALDC black applicants to Harvard had a 4 or worse academic rating,. Non-ALDC applicants with a 4 academic rating had only a 0.02% admit rate. 99.98% were rejected – virtually impossible to be admitted. The admit rate was dropped to exactly 0 for non-ALDCs who received worse than 4 academic – nobody was admitted over the multiyear sample… A 4 or worse academic rating is very well correlated with stats (more so than the better academic ratings), so the majority of non-ALDC Black applicants to Harvard appear to have stats that for all practical purposes guarantee rejection. Most non-ALDC Black applicants to Harvard appear to be wasting their time and cost of application fees. With better knowledge about what stat range has a shot of admission for non-ALDC Black applicants, this could be avoided.