Can AI transform adult online education?

 

By Olina Banerji, EdSurge

Coursera has gained some insights from its 124 million-strong user base about what it really takes to help people succeed in its digital classes. The company employs some high-tech strategies, like a customizable assignments generator that it acquired, for an undisclosed amount, from a Bulgarian startup in 2019.  

“We’ve noticed that the earlier we introduce these assignments into a course, the retention rates improve,” says Jeff Maggioncalda, Coursera CEO. 

Still, completion rates among people who have paid for a Coursera course hover around 50 percent, according to figures shared by the company.

Perhaps some new tools, like Coursera’s newest personal learning assistant, “Coach,” can help change that. Maggioncalda calls Coach a “hands-on, interactive” tool, one that lets learners set their own pace with the material. “Coach is going to be both reactive and proactive for learners. It’s going to be a thinking and writing partner in multiple languages,” he says, typing questions into the chatbot on his iPad. In response, Coach throws up explanations, summarizes lessons, links videos and suggests further courses for the learner to check out. 

Coursera’s tinkering with engagement tools points to a hypothesis about what may be hurtling toward the American higher education system. The next decade could belong to the nontraditional, online learner — but only if the companies and universities that offer remote courses figure out how to ferry such students across the river of distraction and land them safely on the far shore equipped with skills and credentials.   

Demand is out there. It comes from people like Lyndsay Stueve, who works as a full-time global vendor operations expert while raising four kids who are in middle and high school. Stueve started her online learning journey four years ago—first in community college, and then at the University of Florida. She’s now completing an online MBA from Western Governors University (WGU). Stueve’s been an online learner throughout.

Stueve says she needs the flexibility and choice of an online setting. “I like that WGU doesn’t force us into a semester system, with three to four classes every semester. Online, I can choose to go from one class to another, without any time constraint,” she explains. 

But learning online remains a hard nut to crack. Evidence that it works has often been contradictory or disappointing. Today’s online courses are evolved cousins of the early MOOC, or massive open online course. Earlier experiments did not prove as transformative as some advocates hoped, with few students completing the classes they started.  

There are some clear changes in the way online courses are being structured now. Insight about this comes from institutions like the University of Maryland Global Campus (UMGC), a fully online college that has been operational for more than 20 years, which makes it fertile ground to understand how adults really learn online. 

“The adult learner doesn’t care about what 18-year-old, residential students care about. Every minute that we have with an adult learner is a minute they aren’t spending on another priority,” says Greg Fowler, the president of UMGC. Fowler says this realization pushes his team to carefully pick out what to put in each 20-minute video lesson, and how to reinforce that learning quickly.

“We definitely know that we have to make these courses shorter,” agrees Rene Kizilcec, director of the Future of Learning Lab at Cornell University, who’s studied online student behavior closely.

Yet how adults really learn online is difficult to pinpoint, because they are so heterogeneous, says Kizilcec. Different priorities and expectations make one-size-fits-all programming impossible. 

What is clear, however, is that adult learners bring high expectations to online learning. So, for this segment of higher education to grow, companies and colleges will have to figure out how to meet those standards. 

Adult learners aren’t shy about pushing back on course structures that don’t work for them, Fowler says.

“We can’t approach this simply as an authoritarian relationship, where the instructor has the power, and the students just do what the instructor says,” he says. “We get lots of students who raise their hand and say, ‘I don’t think I’m getting what I came here for.’”

A balance of structure and personalization

Chirag Garg, a researcher with IBM, lives in San Francisco and wants to transition to an AI role in a few months. In fact, his company is going to demand new skills from him, so Garg looked for a course that would teach him all the fundamentals of artificial intelligence while being flexible with his work schedule. He landed on Stanford’s “AI Principles and Techniques” online course, and he’s three weeks in. 

“I like how the course topics are sequenced. I’ve done courses before where there wasn’t much of a structure, and I wasn’t motivated to finish them,” says Garg. He also wasn’t paying for those, while such courses at Stanford run at over $1,500 a pop.

The kind of structure that Garg likes isn’t easy to create online.

The first thing that platforms or universities have to do is subvert the linear semester system, and design shorter learning periods instead. “What I’ve observed is that a lot of universities who put the semester system online abandoned it after a while. That’s a crucial change,” says Kizilcec. 

The self-paced nature of these courses is a better fit for adult learners who might have to deal with issues like sickness or job loss. But with self-paced courses, warns Kizilcec, the fear is that the pendulum may swing too much in the other direction—toward no accountability.

Can AI transform adult online education?

It’s a dance that Sourabh Bajaj is familiar with. He’s one-third of the founding trio behind CoRise, a tech upskilling platform that works largely with companies to get their employees up to speed on their technical capabilities. Bajaj is convinced that the flexibility of an online course has to come with some riders. Most online courses have some form of demerits built into them if learners miss too many lectures, or turn in assignments late. CoRise, though, actually makes learners pencil live lectures into their calendars. 

“It creates a cadence. Adults possibly struggle more than younger students to figure out when to study. If you get the option, you’re always going to punt on studying,” Bajaj says. 

CoRise claims to have an 80% completion rate across its courses, and Bajaj boils it down to a tight 8 to 10 hours a week of watching videos and doing assignments. 

“It’s hard to balance a hyper-structured environment with just the right amount of personalization,” Bajaj says.

When it comes to personalization, CoRise is experimenting with both human intervention and AI bots. “Some reminders, information, nudges can be automated,” Bajaj says. “But some problems escalate, where people have to come in and motivate learners. We check in with them at different points to figure out how they are feeling.” 

For Garg, the Stanford student, a human helper doesn’t always seem necessary. He says he often turns to ChatGPT with his doubts, and they are solved on the spot, cutting short the long time it can take to receive feedback in an online course. 

Yet for support with more complex assignments, Garg wants a professor to step in. 

“It’s too much to type into a chatbot. With a human being, I can just screenshot my question,” he says.

Keeping a human connection in online learning

Adults come into the education system at different points in their lives, with different needs. Some are trying higher ed for the first time, while others tried college before but didn’t complete it, and still others have advanced degrees but want training in a specific skill or subject. Online courses have to cater to all that.  

Yet there are some insights that cut across this diversity, Kizilcec explains.

“One intervention that we tried had some of the best short-term effects on engagement. We asked people to find a study buddy, and get them to hold them accountable for their progress. They tell their buddy, ‘I’m going to do this course. Check in with me every week.’ We asked people to do that and plan ahead. We saw that they had more engagement in the course at the beginning,” Kizilcec says.

The study buddy or cohort system means some part of the course has to be synchronous—people logging in at the same time—in largely asynchronous courses. Stanford’s online courses are trying to work around this issue, and faculty have turned their Zoom office hours into a group coaching session. 

“There is a live discussion among learners about how they can apply what they learned in their course to their daily lives. Hearing classmates can validate their own experiences,” says Jennifer Gardner, director of online executive education courses at the Stanford Graduate School of Business.

Dakota Lillie, a current online student with Stanford, thinks the peer connections—fostered through discussions and Slack channels—are a major appeal of the program. “It’s been designed in a way where you can participate [with others] when you want. I like the competition with other students,” Lillie says. 

In contrast, Stueve, at WGU, shudders at the thought of doing another remotely planned group project or discussion. “I’ve been in experiences where I’ve had to pick up the slack in a group project. I don’t really log onto the Slack channels because I don’t need an external accountability partner,” Stueve says.

What Stueve does rely on though, is a personal mentor, who can keep her in sight of her goals. 

At UMGC, Fowler says this kind of support network is something the university is actively building. There is some “peer mentoring,” but a large part of the experience is also delivered by what Fowler calls “success coaches” who reach out if a remote student is struggling in a course. 

“A coach can see that a student got the same question wrong four times in one lecture. And the student may not reach out, but the coach will help them get over it,” Fowler says. 

Without this kind of intervention, online students can simply disappear from courses and institutions, Fowler adds, putting them at risk of becoming another one of the 40 million Americans who have “some college and no degree.”

The future of work and learning together

Innovations in adult learning currently defy clear patterns. Small tweaks—like better feedback systems, study buddies, guides—have indicated a direction, but scaling these services could mean universities and platforms are either shelling out or charging students higher tuition. (Of course, retaining more students can also pay off for programs in the long run.)

Yet if companies and colleges figure out how to help adult students learn, then these institutions may be able to better focus on what adult students learn. And that’s important in a world where evolving technology makes it essential for people to continually refresh their knowledge and skills.  

“We talk a lot about the future of work but not enough about how learning will happen,” Kizilcec says. “There’s going to be a lot of work interspersed with learning.” 

In addition to tinkering with how a course is structured, CoRise has also been putting work into what’s actually being taught. For adult learners, course content should be tightly linked to what is motivating students to enroll, Bajaj says. In many cases, that’s the desire to land a better job, or to gain skills for work. For that reason, CoRise has moved away from relying on knowledge “taxonomies” and toward making content explicitly relevant to online learners, who may not have time for or interest in studying every possible topic. 

Bajaj takes the example of a machine learning (ML) course. “Computer vision isn’t relevant for most companies. Very few companies have image data. But every ML course still has computer vision and it wastes precious learner time. We’re trying to change that,” Bajaj says. Three months into a ML engineer job, you don’t need to know everything, he adds, so getting the relevant skills from an online course is important.

Recognizing micro-skills or issuing microcredentials is another way that course providers are trying to meet adult learners where they are in order to take them where they’re trying to go. Fowler says UMGC is now trying to figure out how existing skills in learners can be “tagged” in a workplace, and if they can be awarded credit for that.

“People are going to need more skills just to keep working. We’re trying to figure out how they can do that without taking time off to do a course,” he says.

Indeed, the learners of the future may zigzig between work and school in a way that might completely transform how online courses are designed. Kizilcec believes that the path to being a nontraditional learner—potentially a majority of students in the future—needs to start earlier. 

“We need to think about how lifelong learners are created at [the] college or high school level. You can’t expect the traditional, residential college to take care of that,” says Kizilcec.

The year of the MOOC may be long over. But the institutions trying to teach new-age learners online are just getting started.

This article was co-published with EdSurge. EdSurge is a nonprofit newsroom that covers education through original journalism and research. Sign up for their newsletters.

Olina Banerji is a writer and reporter based in Washington, D.C., covering edtech, educational innovation, clean energy and health care.

Fast Company

(6)