For partners 
A quiet revolution is unfolding on university campuses, yet many institutions seem unaware that the ground beneath them has already shifted. Students are integrating artificial intelligence into their academic lives, but the response from higher education has been a confusing maze of mixed signals. As one researcher put it, "AI is magnifying the strengths of the strong and the weaknesses of the weak" . The result is a growing education gap—one defined not by access to technology, but by the gulf between student adoption and institutional readiness.
The Student Reality
For today's students, AI is not a futuristic concept; it is a present-day tool. Research from foreign universities reveals that high-achieving students are turning to AI for sophisticated problem-solving, while their less successful peers tend to use it for routine tasks or leisure. In Indonesia, students strategically adopt AI tools, often validating outputs through textbooks or peer input, prioritizing practical benefits despite acknowledging limitations.
But this engagement exists in a policy vacuum. A study of ten selective universities found that just 7.3% reported an outright ban on AI, 6.6% allowed full freedom, and nearly one in five students had no idea what the rules were at all. This opacity—the "mixed messages" that characterize higher education's response—leaves students navigating a minefield of unclear expectations.
Meanwhile, the workplace waits. Employers now expect graduates to arrive with AI proficiency. Mary Adewole, a graduating senior at Miami University, captured the disconnect perfectly: "In many classrooms, AI is stigmatized. However, during my internships, I was strongly encouraged to use it. As a result, I fear that I may struggle to adjust to professional environments" . Her concern reflects a broader structural shift: entry-level hiring is changing as AI assumes many junior tasks, making it harder for new graduates to gain early career experience.
The Country-Level Divide
How nations respond to this challenge varies dramatically, and the approaches reveal much about the future trajectory of their graduates.
In Portugal, a debate rages between prohibition and adaptation. Twenty-eight academics signed a manifesto calling for a ban on generative AI in teaching, warning students are being turned into "digital cretins". The manifesto argues that rapid, uncritical AI adoption has outpaced serious reflection on pedagogy and cognitive development. Yet Portugal's Minister of Education, Fernando Alexandre, has rejected turning back the clock, insisting that "AI is a reality that cannot be ignored" and that adaptation through teacher preparation and curriculum changes is essential.
The United Kingdom faces a different challenge: funding constraints and regulatory burdens. Despite 120 providers offering undergraduate AI provision (up from 47 just a year earlier), experts warn that the UK "risks creating a two-tier graduate workforce—those prepared for AI-enabled careers and those left behind" . Kay Hack, a higher education consultant, notes that "computer science programs are responding robustly to demand, but the broader challenge of ensuring all graduates develop AI literacy relevant to their disciplines remains inconsistently addressed".
The United States exemplifies the mixed messages problem. More than half of students surveyed report that most or all of their instructors currently prohibit AI use outright. Yet employers expect AI fluency. Joseph E. Aoun, president of Northeastern University, argues that "instead of being on the defensive, now is the moment to shake up the way universities prepare students for the world"—and that curricular innovation "cannot be top down, emanating exclusively from the professoriate".
China dominates the research landscape on AI in education, contributing approximately 20.5% of global scholarship. Yet implementation remains fraught. Chinese universities find themselves in a "tug-of-war" between AI essay writers and AI detectors, while parents spend thousands on AI tablets after the government banned human tutors.
Indonesia, despite its large higher education population and rapid digital growth, accounts for only a small fraction of global scholarship on AI in education, highlighting critical gaps in understanding how emerging economies navigate AI integration.
The True Gap: Faculty Readiness
Perhaps the most consequential divide is not between students and institutions, but within the faculty itself. Emma Ransome, a senior lecturer at Birmingham City University, argues that "the real GenAI divide is not between students who use AI and those who do not. It is between academics who understand AI well enough to design learning, assessment and feedback with it in mind, and those who are being asked to manage it without the confidence, knowledge or institutional support to do so".
This unevenness creates inequity. Students whose tutors are confident with AI receive clear expectations, coherent assessment design, and explicit discussion of ethical engagement. Others face blanket bans, vague guidance, or assessments that unintentionally reward undisclosed use. The result is a "postcode lottery" of AI-informed pedagogy, often within the same institution.
What Universities Must Do
The path forward demands a fundamental rethinking of higher education's relationship with AI. Several principles emerge from the global experience:
Redesign assessments, not ban tools. Traditional essays and multiple-choice tests are increasingly irrelevant when a chatbot can complete them in minutes. Forward-thinking educators are designing challenges that build higher-order skills—from data analysis to critical thinking and structured argumentation.
Treat AI literacy as foundational, not optional. Junghwan Kim, an assistant professor at Virginia Tech, identifies three essential components for AI success: technology, domain know-how, and partnerships. Students must understand what AI systems can and cannot do, but technical knowledge alone is insufficient. Domain expertise becomes more important, not less.
Embed AI across disciplines, not just in computer science. STEM students lead in advanced AI adoption, but AI proficiency will be essential for every field. Universities must ensure that humanities and social sciences students develop AI capabilities relevant to their disciplines.
Invest in faculty development. Staff AI literacy cannot remain optional or purely technical. It must be recognized as part of core academic practice, embedded into existing academic processes, and supported through policy developed with academics rather than for them.
Build partnerships with employers and students. Aoun argues that universities must bring employers and students to the table "as full partners in envisioning a curriculum for the AI age" . Employers know precisely what talent they seek, and students—the AI natives—offer insights that faculty alone cannot provide.
Create institutional access, not individual burden. Advanced AI systems often require paid subscriptions, creating inequity based on personal financial capacity. Universities should expand institutional access to advanced AI infrastructure so that proficiency does not depend on who can afford a $20 monthly subscription.
Conclusion
The AI education gap is not about students versus institutions. It is about institutional inertia at a moment when agility is essential. Students are already navigating an AI-infused world, often without guidance, clarity, or consistent expectations. The workplace will demand AI fluency. Universities that fail to adapt risk not only irrelevance but actively disadvantaging their graduates.
The debate between prohibition and embrace is increasingly obsolete. The real question is how universities can prepare students to engage meaningfully with AI—not as passive consumers or fearful skeptics, but as critical, capable practitioners equipped to lead in an AI-transformed world. As one researcher observed, the question of whether AI will level the playing field or deepen the cracks in global education depends less on the technology itself and more on how universities choose to use it.
