Artificial Intelligence & Financial Aid Packaging: Are We Crossing an Ethical Boundary?

Artificial intelligence is generating immense buzz across social chatter. Executives tout its time-saving capabilities that will enhance their workplace’s productivity. Companies of all sizes and scopes are setting expectations that their employees will use artificial intelligence to become more proficient at burdensome tasks that currently reduce productivity. Consumers are engorging themselves on the nifty things artificial intelligence can do for them, such as agentic representatives that handle their mundane tasks or a chatbot that answers questions in a matter of seconds.

Higher education finds itself at the crux of the artificial intelligence debate. Historically, colleges and universities have been slow to adapt to new technologies that improve the delivery of an educational product. In the context of financial aid, many schools still use manual processes when packaging a student’s aid. While automation tools are available in many financial aid administration software systems, some users are reluctant to rely solely on the software. With the introduction of artificial intelligence into higher education information systems, a lively debate is ensuing over the benefits and ethics of using it solely to package a student’s financial aid.

Thirty years ago, most financial aid counselors and aid administrators were manually processing FAFSA forms and aid packages. Some schools relied on paper files to store a student’s sensitive financial aid documents used in verification or other documentation purposes to substantiate a student’s eligibility for financial aid. As one might imagine, this process was overwhelming for the average person employed in the financial aid profession.

Without adequate staffing, financial aid offices were unable to maintain momentum with their institution’s enrollment cycles during peak processing periods. As a result, an opportunity cost arose: failing to hire adequate staffing to run a financial aid operation meant lost tuition revenue for the institution.

Over time, automation tools emerged with the advent of financial aid administration software systems, such as Jenzabar Financial Aid and the College Board’s PowerFAIDS. Automation included integration tools between multiple student information systems to correctly calculate a student’s satisfactory academic progress status, and auto-packaging tools that analyzed a student’s imported ISIR to determine which federal, state, or institutional aid the student may qualify for based on FAFSA results.

Automation tools did not replace human involvement in financial aid packaging. Rather, it enhanced the user experience in the financial aid process by focusing on the larger issues at hand: students who needed assistance completing verification or one-on-one consulting sessions to best plan for the new term. Many aspects of financial aid are extremely complex and require human judgment and input before financial aid packaging can occur. Even the most sophisticated automation tools struggle with aspects of regulatory changes or rules so minute that only the most experienced financial aid administrator can determine how to handle them.

Enrollment management teams are using artificial intelligence to scope out their enrolled classes. One of the more beneficial aspects of artificial intelligence is its ability to perform predictive and prescriptive analytics in real time that would take an average statistician much longer.

For example, artificial intelligence can estimate the likelihood that a student will enroll at an institution based on historical patterns of previously enrolled students. It takes in a large number of data points, including a student’s financial information, financial aid package, demographics, etc., to build a model that predicts which type of student will enroll at the institution. The longer the period of data analysis, the more accurate predictions will become, with the caveat, of course, that the past is not always a terrific predictor of future results.

Prescriptive analytics is a tad different. Instead of using historical data to tell you what may happen, it uses historical data to inform actionable next steps to convert a potential recruit into an enrolled student. For instance, the data may show that a student of this type who receives an SEOG award is more likely to enroll than one who does not. Therefore, you should award SEOG at this amount to increase the odds that students persist as enrolled students. Again, there are no guarantees when a model spits out actionable next steps. Still, it better informs enrollment and financial aid professionals where deficiencies may occur in the admissions funnel.

Chatbots are not necessarily a new thing in higher education. While chatbots have been around for quite some time, especially if you have ever needed to call Comcast or Verizon, their abilities continue to exceed what most of us envisioned 20 years ago. Chatbots can now engage students in conversations about a variety of topics, using a wide range of datasets to inform their output. Think of chatbots as a virtual representative of a college or university. Although most people prefer to speak to a “real person,” chatbots are readily available to answer basic questions, freeing up a real person’s time to address high-level issues that require more time and attention.

Any financial aid professional can attest to the number of documents often needed to verify a student’s eligibility for financial aid. Before automation tools arrived on the scene, financial aid professionals had to manually collate these documents and store them physically in a manila folder in a secure filing cabinet. As financial aid software systems became more nuanced, some began to offer digital storage attached to a student’s record. These automation tools were able to assign touchpoints or other tracking codes to alert students about the verification documents needed to complete the financial aid process. As artificial intelligence evolves, it will be able to scan a student’s financial aid submission, such as FAFSA, and, if applicable, assign tracking codes to the student’s situation while reaching out to the student about their next steps. In many cases, these actions occur before a financial aid professional arrives at the office in the morning.

Other intriguing aspects of artificial intelligence in financial aid include fraud detection and the curation of personalized financial aid experiences. Fraud remains rampant in the financial aid arena, especially with federal student loans and grants. Ghost students are haunting admissions and financial aid offices across the country. Ghost students are students who do not actually exist. Rather, ghost students are fictitious students who use stolen identities to portray real students, and a malicious actor commits financial fraud to obtain a financial aid refund in the student’s name. Before the victim realizes they are a victim of identity theft, a fraudster may have already stolen thousands of ineligible federal loans or grant payments.

Fraudulent activities pose a risk to institutions of higher education because they are subject to strict laws and regulations governing both education and financial institutions. To combat the proliferation of ghost students, some schools are using artificial intelligence software to detect anomalies in a student’s financial aid package that are atypical or akin to a fraudulent financial aid application. Manual reviewers may not always recognize suspicious activity during the financial aid packaging process. Artificial intelligence can play a significant role in combating financial aid fraud, which protects an institution’s image and standing with students and the federal government.

Although artificial intelligence offers several benefits that create a more productive work environment in a financial aid setting, there are a few drawbacks to consider that blur the ethical lines in higher education.

Algorithmic bias is real. When you scroll reels on Instagram or view what TikTok is spitting out at you daily, you’ll notice how quickly it will redirect into a new direction after you commit too much time to browsing a topic. The same can be true of the use of artificial intelligence in the context of financial aid.

Artificial intelligence models use historical data patterns to inform potential outcomes or actions. Large data sets can occasionally contain human error or biases, which negatively affect the validity of the predictions produced by the artificial intelligence model. Relying solely on artificial intelligence to review previous datasets without first analyzing them to identify potential biases can lead to dangerous assumptions. These dangerous assumptions can propagate inequities in financial aid packaging for certain student groups if a model is left unreviewed or unchallenged. Human contact with a data set is essential to avoid compromising the integrity of artificial intelligence’s recommendations.

Anyone who works in higher education at the student level is familiar with the basics of the Family Educational Rights and Privacy Act of 1974. This federal law governs how colleges and universities disseminate sensitive student information. Financial aid data is the most sensitive data because most FAFSA information comes directly from the Internal Revenue Service (IRS). Handling this data conscientiously is important not only for the student but also for the institution’s credibility. Institutions of higher education should be cognizant of the cybersecurity risks associated with using a third-party vendor to support the operations of the financial aid office. It may be equally helpful for colleges and universities to reassure students and the public that their data-handling practices are in a secure environment that limits the risk of inadvertent data breaches. Once again, human judgment is essential when assessing the appropriateness of using artificial intelligence in a financial aid setting.

The most concerning aspect of artificial intelligence is its lack of human emotion. Believe it or not, humans are complex creatures. Our moods can shift significantly depending on the news, weather, and our immediate surroundings. Financial aid is complex. It involves intricate layers of multiple awarding parties, each with varying rules governing how a financial aid decision is made. Most people who do not work in financial aid cannot fathom why a + b does not necessarily mean c.

Emotions often flare up when a financial aid counselor explains a denial of a financial aid appeal. But logic and reasoning can diffuse a once combustible situation into an easygoing affair. When a school relies exclusively on artificial intelligence to shape its incoming student cohort through predictive and prescriptive financial aid modeling, tensions between students and aid administrators will flare up if it is not easy for aid administrators to explain the institution’s awarding decisions. Fin aid administrators should understand the artificial intelligence tools they use to determine financial aid outcomes. A failure to do so will make it appear more likely that students feel the outcomes were due to bad odds or, even worse, an inequitable outcome.

Artificial intelligence has the potential to revolutionize financial aid operations in higher education. But college presidents, leadership teams, and enrollment managers must understand the ethical boundaries that artificial intelligence can bleed into both positively and negatively. Like any tool, when used appropriately, artificial intelligence can elevate a college’s financial aid operations. Used incorrectly, artificial intelligence can cause inadvertent data breaches that violate FERPA protections, perpetuate biases that lead to poor financial aid award decisions, and undermine the human element of the financial aid process that most people experience during their college journey.

I am not opposed to artificial intelligence in financial aid operations. I am encouraging all participants in the financial aid operation process to carefully consider the role of artificial intelligence in it to address the ethical concerns it poses.