To increase students’ chance of graduating on time, some colleges are using business intelligence technology to analyze aggregate student patterns and predict whether individuals will succeed in their intended majors. When the system senses a risk of failure, it alerts college advisors, who may intervene and recommend that a student seek additional support or change their focus.
An estimated 1,400 colleges and universities now use predictive analytics, and early results suggest the tools can help improve graduation rates, writes The Hechinger Report. But some experts are cautioning that those gains may come at the cost of reinforcing systemic inequities for Black and Latinx students. When the technology is used to analyze student achievement and enrollment data, it may recommend that advisors divert a struggling student away from their chosen major toward a less challenging path.
Replicating historic bias?
Because artificial intelligence (AI) systems train on data from past years, they may generate recommendations that reproduce historic bias in educational attainment. Due to myriad socioeconomic factors, Black and Latinx students historically have graduated at lower rates than white students. Even when institutions’ algorithms don’t consider students’ demographic information, other factors more prevalent among low-income students and students of color can lead them to be disproportionately flagged as “high risk” students. Based off the system’s recommendations, advisors may funnel these students into less rigorous majors, which may lead to careers with lower earning potential.
“There is historic bias in higher education, in all of our society,” said Iris Palmer of the New America Foundation. “If we use that past data to predict how students are going to perform in the future, could we be baking some of that bias in?”
If a student is just starting to struggle with a particular academic path—or is even on the brink of succeeding or failing—and is discouraged by their advisor from continuing in that major, the recommendation could become a “self-fulfilling prophecy” that causes the student to give up, Palmer said. This result for one student could then affect future students, as the AI-advisor system reinforces its own results.
Or, creating a ‘win-win situation’?
Still, proponents of using predictive analytics say the visibility gives students time to change course without wasting precious semesters on prerequisites that may be rendered unusable if they are rejected from a competitive program whose application process occurs late in their college career. Students who have to switch majors at that point often incur extra time and expense, threatening their likelihood of completion. And, The Hechinger Report notes, “because of crushing debt, students who drop out are often worse off than if they’d never gone to college.”
Georgia State University, one of the first universities to use a predictive analytics system, credits the technology—in conjunction with a significant increase in student advisors—with helping to increase its graduation rate by 23 percentage points since 2003. Georgia State also has narrowed achievement gaps: graduation rates for the university’s Black, Latinx, first-generation, and Pell-eligible students meet or exceed those for the overall student body.
The number of students graduating with STEM degrees has doubled since Georgia State began using predictive analytics, although The Hechinger Report says that minority students are overrepresented in some less-lucrative career tracks. Noting that more students are graduating from the university’s toughest majors than ever before, Tim Renick, Georgia State University’s senior vice president for student success, says the AI technology has been a boon for both the university and its students.
“These were students who were paying customers who were walking away from the university who now are actually getting their degrees,” Renick said. “So you talk about win-win situations.”