The growing potential of chatbots in higher education
Experts and universities alike see AI technology as a way to reach students 24/7, and remove barriers to access
Higher education institutions are faced with an array of challenges – what sort of teaching arrangements to put in place; how to manage hybrid learning and working environments; and perhaps most crucially, how best to support students as they embark on a new phase in their education journey as a global pandemic plays out.
But just as the pandemic brought about a swift move towards the digital transformation of higher ed, it also led to a raft of innovations which educational institutions can now take advantage of, with chatbots playing a key role in driving both learner and institution success.
In recent years, AI technologies have been empowering universities with new tools to enable and drive student success by personalizing outreach and connecting with students. Among the tools are chatbots, which experts say will be a growing presence on and off campus, offering assistance across the student journey — from recruitment and admissions to onboarding and holistic student support. “A tremendous amount of information actually can be delivered through chatbots,” says Susan Morrow, vice president of product management at Salesforce Education Cloud.
Products like Service Cloud Einstein offer an integrated set of AI technologies including chatbots, which can answer student questions at all hours of the day and night, directing queries to the right department, and directing to in-person support. Crucially, the development of these bots takes the student voice into account, and creates a nuanced dialogue with users, which reflects the diversity of life on campus.
Offering agency — and anonymity
Chatbots offer real-time assistance to today’s students, who are accustomed to such features on commercial websites. And they answer a need. Global research conducted by Salesforce.org on the student experience during the pandemic found that students pinpointed schools’ websites as their primary source for information about the institution before deciding to enroll. The top challenges students identified included maintaining their well-being, and dealing with financial concerns; while many said they would welcome more help with managing course loads and organizing their work and personal life — all of which are opportunities where chatbots can provide added value.
Meanwhile, less than a third (29%) of students said they could easily get their questions answered at their college or university.
By appearing readily on a school’s website, chatbots can address many of these concerns simultaneously. A bot meets students where they are, rather than solely during specific business hours, Morrow explains. “That is both the 24/7 nature of the chatbot but also the reality of being a student and the changes of higher education.”
Furthermore, as Morrow suggests, this technology taps into a “generational specificity.” Students are always on their phones, working or communicating, she says. “Twenty-year-olds use technology differently than twenty-five-year-olds, different than thirty, thirty-five-year-olds. That’s something that’s important to be thinking about — what your customer, your student, is expecting.”
Because the bots offer a sense of anonymity, they are also practical tools in addressing the complexity of the student experience. Some may find it easier to ask a bot questions about topics that may be sensitive or seem embarrassing. Rather than presenting a block of information through which users have to navigate, the bots allow students to make their inquiries, and get a clear answer.
This conversation-first model is what many students now expect, Morrow says. “They get to start that conversation when they want, on whatever topic they want, and drive from there, rather than the more old-fashioned approach of — ‘Here’s a catalogue, read through it and then ask us questions.’”
Building Chatbots with Empathy in Mind
Chatbots were first developed as early as the 1980s, and since then the technology underpinning them has become increasingly sophisticated. Over the past ten years, natural language processing has enabled designers to take into account the way that humans interact in real life. This has “pushed them into a new realm,” according to Greg Bennett, director of conversation design at Salesforce.
Bennett is a sociolinguist whose work draws on ethnography, and his study of how design can cultivate empathy informs his view of chatbots. The goal, he says, is not to mimic human behavior, but to develop a technology that adheres to users’ expectations.
“Because this machine is entering into a process that humans have considered to be exclusively human and have certain expectations around how to navigate with other humans it’s very important—for the sake of building trust with users—to make it clear that they’re talking to a machine and not another human,” he explains.
Designers can do a lot to ensure a bot is sensitive to users, tweaking the timing of its responses and adjusting its language, Bennett says. “Some of those things could include leaving a pause before responding and thinking about the length of the pause we leave before the bot responds. We know from linguistics that the shorter the pause between when the user is done talking and when the bot responds, the more involved in the conversation the bot can come across, or the more that the user could feel hurried by the bot.”
Bennett notes that there should always be an escape hatch — an option for students to talk to a human if they prefer. But if users are upset or angry, the bot’s behavior could be modified accordingly. Something as apparently simple as swapping an exclamation point for a period can alter or soften the tone of an exchange, he says. “Emojis, exclamation points, those are the kinds of signifiers in text-based conversation that can communicate enthusiasm to a user. If the user has an issue, we don’t want to be enthusiastic, we want to show deference.”
Another important area in which chatbots can contribute is by mirroring the diversity of speech on campus. Language can vary greatly, especially in chat and text messages. Bennett points out that a whole field of linguistics is dedicated to studying World Englishes. “Rather than making a judgment we want to celebrate the variety that we see in language, in the model.”
This is why it is crucial for developers, companies and universities alike to take the rich multiplicity of real-life speech into account. To do so, they must engage their users and, in the case of higher education, listen to what students have to say. Or, as Susan Morrow puts it, “If you really want to understand what students need, you let them teach you.”
Listening to students’ voices
Salesforce.org put this principle into action this summer, when it organized a case competition on chatbot design and invited students across the country to take part. It offered introductory sessions to explain how chatbots worked, and then let the students loose on the program. Susan Morrow, who was one of the competition’s organizers and judges, says it was an exciting process. "It offered an opportunity for us to learn. How did they expect to see this? How did they expect these chatbots to be used?”
One of the winning bots was from the University of Washington, called Husky Helper after the school’s mascot. One of the students who developed the bot, Emma Weiss, is a marketing and finance major at the University of Washington’s Foster School of Business and is involved in undergraduate diversity services. The Husky Helper was able to say where a student could find halal, kosher or vegan meals, and how to access the right student-support services.
Diversity and broadening access were central principles underlying the bot’s design, Weiss explains. The team had written more than a thousand utterances — inputs the chatbot will recognize, in the form of different versions of varying questions students could pose. “With that comes variety in both the audience that we’re catering to and the content that we have,” she says. “On the conversational design side of it, it was just really about making sure that we weren’t being exclusive in the language we used or even the spelling or grammar, or making assumptions of what people’s prior level of knowledge is before using the chatbot.”
The Husky Helper catered to both parents and students. Through its platform, among many options, parents could receive alerts about safety, or about how to pay tuition. Students could ask questions about in-person learning arrangements, or pose broader queries about career and life goals.
In their presentation showing how the bot would work, Weiss and her team note that mental health is closely tied with student outcomes. In the second edition of Salesforce.org's Connected Student Report, 76% of students shared that maintaining their well-being is a top challenge; while on campuses like UW, three quarters found it hard to locate disability accommodations, mental health resources or academic help. In big schools there is definitely a place for making it easier for students to obtain information, Weiss explains. “It could really go a long way,” she says. “There are so many questions that can very easily be answered in one spot.”
Making technology work to benefit students
Morrow, whose background is in social justice advocacy, believes passionately that chatbots can improve outcomes for disadvantaged groups. Text-based inputs are easier for students with disabilities to access. By making it more straightforward to get the information students need, the technology is removing barriers, she says.
“Having that safe environment to be able to ask is actually a big part of equity,” she says. “You meet people where they are and you remove those barriers. You make it easier, you level that playing field by providing information equally to everyone.”
There’s another side to this too, she adds. When they do speak to students, university employees already have an extensive understanding of what the issue is. “If a chatbot is used with a capacity for the institution to follow up, then you have a much deeper and personalized relationship.”
Looking forward, chatbots have the potential to further broaden participation in higher education. For Greg Bennett, one of the most promising areas is voice-based assistance. Students are familiar with Alexa, Google Voice and Siri, and voice-activated technology could better support those who are blind. “That would be a natural progression for this kind of technology at the university level,” he says.
Experts observe that the pandemic has shown the positive role for digital-first engagements across higher education. Susan Morrow sees artificial intelligence as a vital tool in helping students meet their goals. “That’s another thing we learned with Covid,” she suggests. “Technology doesn’t fix things, people fix things. Technology is our tool.”
Watch Salesforce.org's on demand webinar to learn how chatbots can deliver holistic student support, from anywhere.
This content was paid for and created by the Salesforce.org. The editorial staff of The Chronicle had no role in its preparation. Find out more about paid content.
Built with Shorthand