Higher education is set to be transformed as AI enables personalized learning experiences on an unprecedented scale
Just as in any industry, higher education has some cardinal rules. One of these theories is the ‘Iron Triangle,’ the concept that there are three primary goals that higher education institutions strive for: accessibility, affordability, and quality. However, achieving all three simultaneously is often believed to be challenging; if one side of the triangle is improved, it may negatively impact the other two. For instance, increasing the quality of education might lead to higher costs, making it less affordable or less accessible to students.
This rule anchors most universities in a traditional model of higher education – one based on quality through scarcity, or at least quality constrained by a scarcity of resources like faculty and classrooms. This scarcity contributes to selectivity, and selectivity is interpreted as the top indicator of quality.
However, this conflation of the selectivity signal with quality is misleading, or at least incomplete. Selectivity is a function of many things, but at its heart is capacity, not quality. Top-ranking institutions, like Duke, turn away very impressive students every year – not because they cannot achieve what the university asks of them, but due to very real capacity limitations.
Yet things are quickly changing. We now know how to overcome some of those capacity problems: through digital learning. Especially over the past decade, leading institutions have devised and honed sound pedagogical approaches for online design and delivery, alongside business models that leverage digital access to education. These models have proven successful at top research institutions, like Georgia Tech, the University of Illinois, and the University of Colorado-Boulder. Even highly selective institutions like Berkeley and Dartmouth recently rolled out master’s degrees that operate with these innovative approaches.
Worldwide, more than 80 universities have cracked the code and broken the Iron Triangle by leveraging technology to scale.
Digital delivery models and learner outcomes
I lead a unit at Duke University focused on learning innovation. Much of our work focuses on the adoption of emerging technologies like generative AI to transform how we learn, teach, work and connect with others, often in profound and permanent ways.
Digital learning is an innovative tool that can address practical educational problems, like the Iron Triangle or other challenges resulting from capacity issues. But digital learning is also the most student-centric approach to education, with a relentless focus on students’ educational needs and outcomes.
Our measures for educational quality are outcomes like retention and completion – they illustrate student success and ROI, and hold us accountable for creating positive change in our students’ lives. This can mean innovative ways to teach and assess learning, and it can also mean spotting challenges and roadblocks, and removing or mitigating them.
To help students meet their goals, it’s essential that we can detect if a student is in trouble, and quickly and effectively provide support. A good portion of this is designing the sensors in the learning environment for detection of such data. I see this as both a function of technology and tools, and of design thinking that pays attention to and invests care in the creation of the learning environment.
The good news is that when students access learning through digital means, they leave lots of digital footprints. Our technology ecosystems are getting more and more sophisticated at capturing these footprints. And new advances in generative AI allow us to better personalize our teaching, mentoring, and tutoring approaches to where, when, and how students need them. Imagine students having their own Socratic teacher, continually asking questions to enable their learning. Imagine allowing students to take as much time as they need to achieve mastery of concepts. Imagine many of the simple or repetitive questions being answered quickly, freeing the instructors to do the more significant, more humane, and more transformative work with their students.
Then imagine being able to adopt that approach at scale, for large groups of students, at a relatively marginal cost per student. These environments are already being created – and they will improve over time.
Global problems and scalable solutions
Today’s problems demand solutions at scale, not boutique or artisanal approaches. There are countless opportunities for generative AI technologies to improve our learning, our work, and our life – and to do so at scale. In education, this looks like the creation of truly personalized learning environments for a large number of learners that maximize each learner’s individual potential. AI has hugely exciting potential for providing scalable solutions for education and workforce development at a time when we know that rapid changes in technology, automation, and global markets will require education across a lifetime – not only through traditional models like degrees, but also through alternative models and credentials like professional certificates and massive open online courses.
Powerful though AI is, it is crucial for us to extend beyond the technical, delving into the ethical implications of AI and its environmental footprint. It is also our responsibility to think deeply about the profound influence of AI on our understanding of knowledge, and how AI transforms human agency. Being a leader in the age of AI is about balancing our excitement about the potential of these technologies with great concern over the trustworthiness of AI models, AI’s climate impact, and the inevitable inequities in access to these technologies, which we will have to address.
Generative AI will change so much of what we do and how we work. In education, as in other industries, we need to leverage this technology and its ability to scale in order to build highways, not promenades.
Dr Yakut Gazi is vice provost for learning innovation and digital education at Duke University