What is AI?
The section that follows addresses the application of Al to solve three major issues confronted by higher education. However, it is first necessary to explain what Al is and what it is not. Though there is some argument among purists, Al is a broad commonly used category that encompasses artificial intelligence (Al), machine learning (ML), and deep learning. In simplest terms, they are actually successive stages of computer automation and analytics that are built on a common platform.
On the first tier of this platform sits Al, which analyzes data and quickly delivers analytical outcomes to users. Machine learning sits on the tier two application of Al, which not only analyzes raw data but also looks for patterns in the data that can yield further insights. Deep learning is a third-tier application, which analyzes data and data patterns but goes even further. The computer also uses advanced algorithms developed by data scientists that ask more questions about the data with the ability to yield even more insights. The following is an excerpt from an article by Shacklett (2019) that provides insight as to how these three techniques work together.
Tier 1: AI
You develop an Al application that tells your traffic engineers and planners where the major traffic congestion points are located in the city.This assists them in planning for road repairs, stop lights, and other infrastructure that, hopefully, can relieve congestion in certain areas.
Tier 2: Machine Learning
You further develop your Al/analytics so that it also looks for patterns in the data. For instance, it notices the traffic at certain intersections is most congested in the morning between 6 am and 8 am, or that traffic queues up in the evening, ahead of a sporting event.
Knowledge of the situation gives planners and engineers more insight because now they can plan not only for traffic snarls but also for future events like concerts and hockey games.
Tier 3: Deep Learning
Deep learning is where data analytics moves beyond raw data and data patterns. Deep learning adds specific algorithms that data scientists develop to further expand the querying and insights derived from the data.
Algorithms that could be added to the traffic analysis might include: What areas of the city will see the greatest population growth over the next ten years? Or, which roads will need major repairs in the next five years? Or, do weather projections say that we will have more or less snow over the next five years? By adding these algorithms on top of pattern and data analyses, users get a more complete picture of the situation they are trying to act on and assess.
Beginning the Student Journey with Help from AI
If we look at the student journey, it is apparent that costs begin to escalate even before students enter the institution. In a recent Noel Levitz report (2018), the cost of taking a student from prospect to enrollment runs between $714 from students at public four-year institutions to $2,537 at private universities. To understand the impact of this metric, consider a public university in which 10,000 new students are enrolled per year. In this instance, the total cost of acquisition is $7,140,000 before the student ever takes their first class.
Colleges and universities are under increasing pressure to make their enrollment numbers and are increasing their costs to generate that class. The issue is two-fold. On the one hand, in certain parts of the country, college-aged student demographics are down, and on the other, students, armed with the ease of applying to multiple schools thanks to the Common Application, are applying to more colleges than ever. Outside of the most select schools, colleges are struggling to know who and how many students will actually apply. That has led to increasing costs to attract students. According to education research and consulting firm, Gray Reports (Simon, 2019), the Price-per-Inquiry at colleges and universities had increased year-over-year in 16 of the 17 months leading up to August 2018.
For selective institutions, this means potentially spending more than necessary to recruit new applicants and giving away valuable space to students who may not retain. For non-selective institutions, there is significant potential for large numbers of underprepared students to attrite if they do not receive the proper types of support immediately.
A potential solution to this issue can be seen in actions taken by Laureate International Universities (Laureate) in Latin America. Laureate is the most extensive global network of degree-granting higher education institutions, spanning the Americas, Europe, Africa, Asia, and the Middle East. The network consists of 60 campus-based and online universities in 15+ countries offering undergraduate and graduate degree programs to over one million students worldwide, including undergraduate, masters, and doctoral degree programs in fields such as business and management, medical and health sciences, engineering, information technology, architecture, education, law, communications, and hospitality management. Every institution in the Laureate network operates as its own unique brand, guided by local leadership, and actively engaged in its community, enriched by shared curricula, faculty, degree programs, and student exchange opportunities. As with other institutions, Laureate saw the increasing costs associated with admissions as one barrier to their ability to offer affordable tuition.
To solve this problem, they turned to a regional analytics company, Analytikus, who they had worked with on other initiatives, and Microsoft to devise a solution. Starting with a deployment at UNITEC in Mexico, the companies used Al technologies, along with traditional data science techniques, to create a solution that helped the university understand which prospects have the highest value in terms of likelihood to convert to enrolled status and then persisting through to graduation. Additionally, the Al engine helped surface appropriate next steps for admissions representatives to take when interacting with prospects, to streamline the process and increase efficiencies. Within one year of implementation, UNITEC’s prospect conversion rate had improved by 10 percent, with commensurate improvements in return on investment. As the technology was refined, second-year results showed a 16 percent improvement over the yearly average before solution implementation. Since that point, Laureate has moved to deploy the solution across more than 20 of their institutions in Latin America, Spain, and Malaysia. Likewise, other institutions in the regions have taken note of the efficacy of using Al to help solve this problem, with deployments at over a dozen other Latin American Institutions.
