Artificial Intelligence in Higher Education: An Inevitable Future?
Have you ever thought of taking a class with an Artificial Intelligence (AI) professor instead of a human teaching the class? Probably not. How about a class with a virtual teaching assistant that could answer your questions about the course content? Again, probably not. These scenarios seem too far in the future to consider right now. However, the latter is already a reality.
In a class named “Knowledge Based Artificial Intelligence” at Georgia Institute of Technology, one of the teaching assistants was Jill Watson. Jill was tasked with responding to students’ questions posted on an online message board for the course. Typically, students posted approximately 10,000 messages each semester. That is typically too much work for any one teaching assistant (TA), but Jill nailed it, answering 97% of inquiries successfully.
Jill’s secret? That she is not a real human but an AI teaching assistant who was trained on data involving student questions and answers from previous semesters. The last name of Watson refers to IBM Watson, a program that offers AI solutions for businesses.
Although Jill was limited to this task and not taught to grade assignments, AI has potential to be used for this purpose in the future.
McCormick junior Vincent Wang believes that using virtual TA’s for grading could work well. “If there’s a virtual TA to help us grade the exams or homework, that wouldn’t be bad,” Wang said.
He added that it is important to “make sure that every student has a right to point out if the virtual TA has done something wrong.”
McCormick junior Nastya Lantsova said she feels that virtual TA’s can be used for grading technical assignments such as math or economics homework, but not more subjective, essay-based assignments.
“It intuitively feels wrong to even imagine that an AI can assess an opinion in an essay as right or wrong,” Lantsova said.
Another use of AI in higher education is by Instructure, a company that owns and operates Canvas, the education management platform used by Northwestern professors and students for online exams, digital material distribution and course-wide announcements. In 2019, Instructure tested Nudge, a tool that “uses machine learning techniques to prompt students to log into Canvas or turn in assignments” when they are not likely to keep up with their courses.
Lantsova believes that such tools can help her and other students to “not slack on courses and do the work.” She said that it is very helpful for students to effectively digest the course content and learn, especially for asynchronous online classes.
AI can be used not only for instruction in higher education but also for other purposes, such as student retention. Georgia State University used an AI chatbot named Pounce to answer more than 200,000 questions from students and reduced the “summer melt” (students choosing not to show up to classes in the fall) by 22% in 2016.
Pounce particularly helped students who faced enrollment obstacles like financial aid or registration issues. Among these students, many came from low-income and first-generation backgrounds, and Pounce allowed them to gain necessary assistance and guidance in the enrollment process.
At this stage of development, AI use in higher education can facilitate learning through the use of virtual TA’s, and it can assist students with administrative issues. But can it fully replace a human professor in the classroom?
SESP freshman Haneef Usmani said he thinks it is impossible right now because students value personal connections with professors.
“We are very far from AI replacing professors. AI is unable to provide the personal touch and feedback that a human professor can give to their students,” Usmani said.
Wang feels that although AI “shouldn’t take the professor’s job,” he would take a class with an AI professor if given the opportunity. He added that the content for such a course should be technical without any projects that involve design, research or essays, as “it wouldn’t be the same without a real person to talk to.”
Thus, if a fully automated learning system with AI replacing human professors is improbable and too far ahead, what lies for the future of AI in higher education?
Wang believes that AI’s primary contribution to education will come through “the capacity of AI for data storage and processing,” which can assist professors with better delivering their expertise on a subject.
An important implication regarding AI’s future in higher education is AI’s susceptibility to biases, which has to be addressed to make learning equitable and accessible for everyone.
Usmani said that it is crucial to “have a diverse team of individuals developing the AI and run it through multiple stages of testing.”
He believes that these are necessary steps to take to address and prevent biases in education that have arisen in the application of AI in other industries.
In all, while AI has recently made innovations in higher education, it still has a long road to go. It is up to us as a university community to decide on and chart the AI’s path moving forward.
