Generative AI and Global Education

Artificial intelligence (AI) is a rapidly developing technology that has the potential to transform education on a global scale. AI can enhance learning outcomes, reduce barriers to access, and foster innovation and inclusion in education. However, AI also poses significant challenges and risks, such as ethical, legal, and social implications, that require careful consideration and regulation.

AI is not a threat nor a competitor. It is a partner and a catalyst. AI can enhance human capacities and empower human rights. AI can transform education for the better if we use it wisely and responsibly. AI can make education more accessible, engaging, and effective, for everyone, everywhere. This is the promise of AI for education.

If that description gets a bit discursive towards the end, that’s because it was written by ChatGPT. And while this, like many of the uses of generative AI that have cluttered our social media feeds, seems trivial, the technology appears poised to take another leap—one with significant implications for international education and global relationships more broadly. It’s not surprising, then, that international educators are trying to understand how the technology could help support students in meaningful ways—and they’re not the only ones wondering how best to integrate AI into their work.

"This is an exciting and confusing time, and if you haven’t figured out how to make the best use of AI yet, you are not alone."—Bill Gates

Just ask Bill Gates. In his year-end blog post, the Microsoft founder wrote, “2023 marked the first time I used artificial intelligence for work and other serious reasons, not just to mess around and create parody song lyrics for my friends.” He continued, “This is an exciting and confusing time, and if you haven’t figured out how to make the best use of AI yet, you are not alone.”

From Turing Tests to Hallucinations

AI isn’t new. The concept dates back to the 1950s, when Alan Turing predicted that computers would become “thinking machines” and posited his eponymous test to determine if their cognition was comparable to that of human beings. Programmers later taught computers to beat the best human minds at chess and, more recently, the even more complex game of Go.

All kinds of algorithms and chatbots have been put into service to automate repetitive tasks. But the emergence of “generative AI”—which has the ability to spontaneously create text, images, and video in response to user prompts—took the world by storm when OpenAI launched ChatGPT at the end of November 2022. Within two months, the service attracted 100 million users, arguably making it the fastest-growing digital tool in history, and it was rapidly followed by similar offerings from Google, Microsoft, and others.

While its ability to respond to even the most obscure questions on the fly appears astonishing, generative AI isn’t magic. Like all computer programs, it takes inputs—in this case, massive volumes of text and algorithms known as large language models (LLMs)—and uses them to generate outputs, the spontaneous responses that have captured the public’s imagination.

Generative AI is now being infused into software for virtually every field and discipline. It’s also being touted as a way to automate more sophisticated processes, such as through more refined document-processing algorithms or less easily confounded chatbots that would help organizations separate routine cases from more complex ones for humans to resolve.

But just as the concept of AI isn’t new, neither are the concerns surrounding it—think of the out-of-control computer HAL in the 1968 film 2001: A Space Odyssey. Two decades beyond that film’s alternative timeline, the term “hallucination” took on new meaning in 2023, as users quickly learned that generative AI will confidently state incorrect information as fact, infuse implicit biases, or, in some cases, make things up as it goes along.

This is in part because the output created by generative AI is based more on probabilities than actual cognition, but it’s also hard to discount the idea that the human-made writing fed into LLMs isn’t all that different from the humans who created it—contradictory, inaccurate, and reflective of the biases and uncertainty that imbue actual thought.

For education, the immediate fear was the impact of generative AI on academic integrity, from students cheating on assignments to researchers incorporating plagiarized or incorrect information. Many school districts, colleges, and universities quickly established bans on the use of ChatGPT and similar tools.

"The debate is much larger than whether the use of generative AI should constitute cheating. It was about harnessing a new skill set for larger academic objectives."—Christopher Alan Bracey

However, early research seems to suggest that AI hasn’t dramatically changed students’ (unfortunately high) propensity to cheat, which among a survey of high school students conducted by Stanford University remained relatively unchanged at between 60 and 70 percent; a separate Pew Research Center study of high school students found that only 13 percent had used ChatGPT to help with their assignments. Now, many colleges and universities are moving beyond creating reflexive bans to actively embracing AI—both as a field of study and a tool to better serve their missions.

“Our starting point was to flip the conversation,” George Washington University Provost Christopher Alan Bracey said last spring. “The debate is much larger than whether the use of generative AI should constitute cheating. It was about harnessing a new skill set for larger academic objectives.”

AI and Higher Education: A $6 Billion Market

Education has been considered one of the primary use cases for AI for more than four decades. The International Artificial Intelligence in Education Society (IAIED), an interdisciplinary community that publishes a scholarly journal on the use of AI in education, was established in 1997 and now has members in 40 nations. But there’s growing consensus that generative AI represents a leap forward in how technology can support learning in a broad variety of ways.

Consider Khan Academy, which added an AI-powered tutor and teaching assistant called Khanmigo to its learning tools and resources last year. Noting that research has consistently shown the benefits of one-on-one tutoring in learning, Khan Academy founder Sal Khan describes Khanmigo as a scalable “above average or good tutor” that follows “pedagogically sound” approaches like asking students to explain their reasoning instead of simply providing the correct answer.

Khan Academy is working on a similar agent focused on writing college admissions essays, with an emphasis on “doing it with you, not for you,” Khan said in an Aspen Ideas podcast, calling it “a major leveler of the playing field.”

“People can pay thousands of dollars for consultants,” he said. “Now every student can have expert coaching.”

Education-related AI is expected to become a $6 billion market by 2025. As it continues to evolve, the technology’s use cases in education extend beyond tutoring and writing support for students. They include:

• personalized learning to meet each student’s areas of need, with predictive analytics that focus content on individual strengths and weaknesses and adaptive testing that adjusts questions to provide educators with a better understanding of student learning;
• real-time support for students taking remote and hybrid courses, with AI-powered chatbots and coaches that provide answers to questions during asynchronous learning;
• chatbots to scale and supplement advising and other in-person student services;
• nonacademic supports for students, including social and emotional supports, mental health resources, and the use of analytics to identify learning disabilities;
• tools for educators to create and enhance existing courses. Coursera’s AI-powered tools, for example, prompt professors to outline objectives and then help bring together course materials. “It went from a process that could take months to a few minutes,” says Jesús Rosario, the company’s senior partnerships manager.

Many colleges and universities are also beginning to intentionally introduce students to AI to prepare them for careers after graduation. New jobs with titles like “prompt engineer” and “language model trainer” have exploded with the proliferation of generative AI, and research firm Forrester says 60 percent of all workers will use AI in their jobs in 2024.

For these and other reasons, it’s no surprise that higher education is poised for change. A recent Educause survey of higher education stakeholders found that 83 percent believe that “generative AI will profoundly change higher education in the next three to five years.” Nearly two-thirds (65 percent) also believe its use “has more benefits than drawbacks."

“AI represents a massive opportunity for us to help level the playing field for a lot of people when it comes to access, equity, and affordability."—Jesús Rosario

And while many of the use cases for AI are being explored at an institution-wide level, the technology could dramatically reshape international education—both at U.S. institutions and those around the world.

“AI represents a massive opportunity for us to help level the playing field for a lot of people when it comes to access, equity, and affordability,” Rosario says. “We shouldn’t bury our heads in the sand.”

Bringing the World Together

AI supporters both inside and outside of higher education point to the technology’s potential to serve the global population of students more equitably.

“One of the things that excites me the most about this type of technology is the possibility of localizing it to every student, no matter where they live,” Gates wrote in his blog post. He points to an AI-based tutor called Somanasi being developed in Kenya: “The name means ‘learn together’ in Swahili, and the tutor has been designed with the cultural context in mind so it feels familiar to the students who use it.”

“It’s really going to change the landscape of what we do.”—Jesús Rosario

For Coursera, whose online courses serve 130 million learners globally, generative AI allowed the company to increase the number of courses it translates into multiple languages from 200 a year to more than 2,000 in a quarter, according to Rosario. “That’s a massive increase in accessibility,” he says. “It’s really going to change the landscape of what we do.”

College students in other regions of the world are also more familiar with generative AI tools. One survey conducted last year by Anthology suggested that college students from outside the United States are more than twice as likely to use AI tools frequently—23 percent versus 10 percent. And expectations matter. More than half (52 percent) of global students believe AI will “revolutionize teaching and learning,” according to the survey.

The Impact on International Education

For international offices, AI can both help bring students to campus and support them once they’ve arrived. For example, the same tools domestic admissions offices are exploring can help market institutions to international students in targeted countries and regions. “The current process of preparing content for specific regions is a lengthy process,” Rosario says. “AI can be leveraged to simplify and scale the process—and do a better job of it.”

AI-powered chatbots, translations, and other supports can then guide students through the application process, providing answers at times when human staff aren’t available. Ocelot, a company that provides AI-powered text messaging and chatbots to nearly 500 universities, notes that nearly 40 percent of all interactions happen outside of business hours—and that’s among users who aren’t many time zones away from institutions, as international students often are.

“International education pros need more support, and AI has shown to be a valuable way to address resource challenges."—Erik Larson

ChatGPT and similar tools can be trained on an institution’s own information to provide expert responses to procedural questions, such as application and visa requirements. Further down the line, avatars could add a human face—and real-time translation—to these automated responses. And AI-powered tools could manage the internal pieces of complex processes, including visa processing and distributing orientation materials. Doing so “helps office staff become more strategic instead of answering generic questions a million times a day,” Rosario says.

Bespoke tools are beginning to be created for international education. Global Study and Terra Dotta, for example, both announced last year their intention to integrate generative AI into their international education products.

“International education pros need more support, and AI has shown to be a valuable way to address resource challenges,” said Erik Larson, Terra Dotta’s chief technology officer, at NAFSA’s annual conference in 2023.

Leveling the Playing Field

Once international students are on campus, research has shown that they face common academic challenges, including language proficiency, understanding instructions, completing assignments, conducting research, participating in discussions, and keeping pace with the rest of the class, according to a paper published in Applied Sciences by U.S., Italian, and Greek researchers in fields including education, computer science, and economics.

“Consider an international student navigating a new academic environment in a foreign country while balancing cultural and language differences,” the paper’s authors state. “With these added challenges, the student needs access to resources that can help them succeed academically.”

Existing, readily available tools such as Grammarly and Google Translate are adding AI-powered features to ease communication. Researchers have found that ChatGPT was able to pass English reading comprehension exams in the Netherlands, earning grades comparable to the mean grade of all Dutch students who took the exam. And AI-powered chatbots could provide even greater support—studies have shown they can increase students’ motivation, self-confidence, and interest in learning English. “[These kinds of tools] bring a lot of opportunity to level the field,” Rosario says.

And when imbued with global context, personalized learning tools may help bridge cultural differences in learning. “Consider two learners: one is an international student from India and the other is a domestic student at a U.S. university,” the Applied Sciences paper states. “Both could receive assignments on a common topic, such as the political theory of democracy, but with more relevant examples of their background and experience. Democracies may look different in different countries, and providing such personalized learning experiences can enhance their understanding.”

AI-powered learning analytics could also help identify factors “that contribute to the success of international students” to develop “more effective support programs tailored to international students’ needs,” the paper states.

In keeping with AI’s potential to level the playing field for learners, the technology could also help support another international education practice with a similar goal: virtual exchanges, which provide more accessible global experiences but have proven challenging to scale. “A lot of the challenges involve content creation, translation, and language,” Rosario says. “Execution is where people are getting tied up—the faculty time that works to create an experience that works for students on both sides. Without that, it doesn’t level the field.”

But the limitations of generative AI could also create significant challenges. Along with data privacy issues and the potential of perpetuating long-standing biases, “AI-powered voice recognition and dictation tools may struggle with accents and dialects, potentially leading to misunderstandings or miscommunications,” the Applied Sciences paper cautions. “Additionally, AI-based English-language learning applications may not offer a broad enough view of language and culture, limiting students’ exposure to diverse perspectives.”

“You always have to have someone behind [AI]."—Jesús Rosario

Rosario notes that generative AI has one important capability to address these challenges that typical software and analytic tools do not. “AI is not static,” he says. “It can be trained when you take the experience of the learners and feed it back into the model.” But doing so requires a commitment to set guidelines for using and monitoring the technology. “You always have to have someone behind it,” he says.

What’s Next?

The one thing AI experts seem confident about is that the technology will continue to grow more powerful in the years to come. Whether it’s called AGI (artificial general intelligence) or “frontier AI,” the technology is expected to progress beyond specific tasks (so-called “narrow AI”) to much more complex, and humanlike, ones.

The impact on global education—and the state of international relations beyond higher education—could be profound. International educators have focused on the impact of authoritarian governments around the world in recent years, and evidence has already surfaced about state actors using AI-generated photos and other materials to spread disinformation. “I worry about what’s going to happen with authoritarian governments monitoring their people with generative AI,” Khan said in the Aspen Ideas podcast.

“The promise of ‘AI for all’ must be that everyone can take advantage of the technological revolution underway and access its fruits, notably in terms of innovation and knowledge.”—UNESCO

The technology may also already be exacerbating existing divides. For example, some of the work of cataloging and sifting harmful materials out of the data used to train generative AI models has been outsourced to low-paid workers in the Global South.

UNESCO, which has focused extensively on AI in education, “aims to shift the conversation to include AI’s role in addressing current inequalities regarding access to knowledge, research, and the diversity of cultural expressions and to ensure AI does not widen the technological divides within and between countries,” the organization’s website states. “The promise of ‘AI for all’ must be that everyone can take advantage of the technological revolution underway and access its fruits, notably in terms of innovation and knowledge.”

And educators have a key role to play in meeting those goals. “If we can work on the core issue of human potential and human intelligence using AI, then I have a lot more confidence that we as a society are going to be able to tackle all the other issues that AI is going to throw at us,” Khan said.  •