right arrow
Examples: Monday, today, last week, Mar 26, 3/26/04

Are AI-assisted college admissions on the horizon?

Replies to: Are AI-assisted college admissions on the horizon?

  • Groundwork2022Groundwork2022 2226 replies37 threadsRegistered User Senior Member
    Well if colleges were smart - and they are - they'd already be using programs to flag applicants who don't meet minimum GPA or test score requirements or who have particularly high scores. It improves the efficiency of the review process.
    · Reply · Share
  • Techno13Techno13 175 replies7 threadsRegistered User Junior Member
    @Groundwork2022 true (and larger schools probably are) but that's not AI. I would HATE to see people removed from the process, but AI-assisted implied human involvement/final decision. Still, it reeks of every dystopic novel I've ever read.
    · Reply · Share
  • AlwaysMovingAlwaysMoving 163 replies1 threadsRegistered User Junior Member
    Schools already have a computer read the applications and provide feedback. The level of the feedback depends on the school, but it has to be a bigger and more nuanced impact at highly selective schools.

    It's similar to a job screening at a top tech company. My wife was involved in one, and they had over 1000 resumes that were qualified. There is no practical way for my wife to read 1000 resumes, so the AI program gave her 24 resumes that were the best of 4 different types of candidates.

    How could a school like Stanford not do something similar?
    · Reply · Share
  • ucbalumnusucbalumnus 78232 replies690 threadsRegistered User Senior Member
    AI could theoretically make the process more repeatable and less subject to random variations due to an applicant getting a human reader having a bad day or whatever. It may also give the admissions staff the possibility of doing multiple "what if" scenarios giving different weights to different application characteristics (GPA, test scores, extracurricular type, extracurricular achievement, etc.) for the same applicant pool and then choose the admit pool they like best.

    But yes, there are reminders of dystopic future scenarios...
    · Reply · Share
  • Mom2aphysicsgeekMom2aphysicsgeek 4577 replies56 threadsRegistered User Senior Member
    There is no doubt that the SRAR is processing student data and assigning students to categories (presumably review/reject). It wouldn't surprise me at all if in the near future that even greater filtering and narrowing of applicants takes place before possible candidates' info hits an ad coms desk.

    Surreal? My Dd and I just had a conversation about Fahrenheit 451 and the focus on the talking walls vs talking to people. So smartphones aren't walls......
    · Reply · Share
  • rickle1rickle1 1935 replies17 threadsRegistered User Senior Member
    AI and holistic, hmmm.... I know folks will say AI will be used to screen to a certain round and then holistic kicks in. Not really holistic IMO, but I don't see an reasonable alternative when the common app has increased app counts dramatically.
    · Reply · Share
  • Mwfan1921Mwfan1921 2243 replies30 threadsRegistered User Senior Member
    edited September 25
    There is no doubt that the SRAR is processing student data and assigning students to categories (presumably review/reject). It wouldn't surprise me at all if in the near future that even greater filtering and narrowing of applicants takes place before possible candidates' info hits an ad coms desk.

    Some schools' predictive analytics models calculate the likelihood each student will attend and places those apps in deciles before any AO sees the file. The variables and weighting of the variables are unique to each school, and based on prior years' experience showing which factors influence whether a student accepts an offer of admission....these factors include things like travel distance for a visit, other types of demonstrated interest, test scores, GPA, rigor, AP courses, etc. etc. Applications in deciles that suggest they won't attend if offered admission do not necessarily receive the same level of review.

    None of this is AI, but interesting none the less. The number of schools using advanced modeling techniques for admission/enrollment management is increasing every year.
    edited September 25
    · Reply · Share
  • ucbalumnusucbalumnus 78232 replies690 threadsRegistered User Senior Member
    Mwfan1921 wrote: »
    Some schools' predictive analytics models calculate the likelihood each student will attend and places those apps in deciles before any AO sees the file. The variables and weighting of the variables are unique to each school, and based on prior years' experience showing which factors influence whether a student accepts an offer of admission....these factors include things like travel distance for a visit, other types of demonstrated interest, test scores, GPA, rigor, AP courses, etc. etc. Applications in deciles that suggest they won't attend if offered admission do not necessarily receive the same level of review.

    Estimating applicants' or admits' likelihood of matriculating is not new. Some of the more obvious factors that have presumably long been considered:

    * ED => almost certain to matriculate.
    * Other level of applicant's interest => correlated to likelihood of matriculating, but can be gamed by applicants.
    * Stronger applicant => less likely to matriculate due to having more other options.
    * Cost / financial aid / scholarships => college may be competing at least partially based on cost against other colleges (or not attending college at all).
    · Reply · Share
  • Mwfan1921Mwfan1921 2243 replies30 threadsRegistered User Senior Member

    Estimating applicants' or admits' likelihood of matriculating is not new. Some of the more obvious factors that have presumably long been considered:

    * ED => almost certain to matriculate.
    * Other level of applicant's interest => correlated to likelihood of matriculating, but can be gamed by applicants.
    * Stronger applicant => less likely to matriculate due to having more other options.
    * Cost / financial aid / scholarships => college may be competing at least partially based on cost against other colleges (or not attending college at all).

    I didn't say it was new.....just that use is increasing and the models becoming more detailed and complex. Some of the models have dozens of variables in them and are re-run every night to project yield, revenue, etc. as the admissions committees move applicants in and out of the admit pile. Paul Tough's new book has a chapter that highlights the process at Trinity College (CT).

    · Reply · Share
Sign In or Register to comment.

Recent Activity