ChatGPT and the emerging generative artificial intelligence offerings from Google and Baidu have made AI the hottest topic in higher education. Whether it’s poised to doom the academic essay or not remains to be seen (to bend the Mark Twain quote, rumors of its death are probably exaggerated). But one thing is clear: AI is here to stay, and it may one day play as big a role in teaching, learning, and the services supporting them as the web does today.
While the notion of AI being capable of taking exams with passing grades at prestigious law and business schools may be new, the idea of using AI to as a tool to improve higher education is not. The International AIEd Society, which publishes the International Journal of AI in Education (IJAIED), was founded back in 1997. More recently, a 2019 report in the International Journal of Educational Technology in Higher Education reviewed 146 articles about AI-related research at universities and came up with 17 categories. Among them:
- Student profiling and prediction of dropout and retention
- Admissions decisions and course scheduling
- Identification of low-engaged students for early intervention
- Intelligent tutoring
- Teaching course content
- Supporting teachers in course design
- Diagnosing strengths or gaps in student knowledge and providing automated feedback
- Curating learning materials based on student needs
- Automated grading (this, we have learned, can be fraught)
Four years hence, some of these ideas are further along than others, and new applications for AI in higher-ed have emerged. The most promising of them share the irony that AI-enabled depersonalization can make the student’s – or the teacher’s – experiences more personalized. That’s because teaching, counseling, administrative, and other workloads often make the alternative to AI-driven personalization no personalization at all.
Advise, AI
So, besides ChatGPT-generated essays (and computer code), what AI-powered solutions are we seeing at universities? One interesting example is that of a prestigious U.S. business school harnessing AI to improve the student-advising process. The system allows the visualization of relationships between classes, activities, the student’s academic history, and the academic backgrounds and career moves of past students to create personalized recommendations on courses and other activities.
This process doesn’t eliminate human academic advisors – quite the opposite. Advisors use the system’s dashboard during meetings with students, providing an individualized depth and thoroughness otherwise impossible given the school’s quite typical 150:1 student-to-advisor ratio. Often, the AI advisor identifies opportunities for student engagement that might never have otherwise surfaced.
The department of philosophy at a German university is using AI in a different advisory capacity. Geared toward prospective students or those enrolled elsewhere at the university but curious about philosophy, the department created an AI chatbot to answer such questions as, “Why should I study philosophy?” The philosophy behind the idea was to provide a 24/7 resource for lots of students asking lots of philosophical questions, lend a modern flair to an ancient discipline, and attract students to the field.
AI is also coming into play in academic research. While AI is already a mainstay in advancing academic fields such as physics and chemistry, it’s now smoothing the path for establishing research projects in the first place. AI-powered solutions can help investigators identify what similar (or identical) research projects are already out there and then narrow the universe of possible funding sources. In this case, AI is lightening the load of vital yet labor-intensive work that brilliant scientists would prefer to not spend their time on.
AI is also showing promise on the development side of the house. Based on donor and prospect profiles, giving history, and other inputs, AI is helping universities are better understand who their donors and prospects are. That enables more personalized pitches and incentives that can to bring in bigger dollars for education and research.
Scratching the surface
There are, of course, challenges: privacy considerations, ethical considerations, pedagogical considerations, and, not least, financial considerations. University budgets are tight, and machine-learning talent is scarce and well-paid. Penn State University has in part addressed the talent challenge by employing interns and new graduates from its own computer science programs. Their work has already automated aspects of IT testing to cut human time spent by 80%, and it’s poised to rationalize mailroom operations.
These examples are just scratching the surface of what AI is doing for universities today, and there’s so much more to come. But this sampling tells an important story. It’s that effectively harnessing AI in higher education is about considering processes – learning processes, teaching processes, and business processes – and thinking hard how machine learning can improve, streamline, and partially or fully automate them.
It’s no coincidence that none of the above examples are university-wide initiatives. Starting with a discrete problem at the college or department level lowers the stakes and provides the leeway to turn projects more quickly, evaluate and learn them more easily, and notch successes that help justify new AI initiatives. It’s early days for AI in higher education, after all, and we’re still learning about a powerful tool we rightly sense will have a profound – and, I think, profoundly positive – influence on university education and research for decades to come.
Malcolm Woodfield Global Vice President and Head of the Education & Research business unit at SAP.
For more articles on ChatGPT see:
Professors Face Challenges with Students Using AI Writing Tool