I asked three large language models to tell me what jobs in higher education are most vulnerable to replacement by artificial intelligence in the next five years. I also asked for recommendations for current and prospective employees in the field.
First, let me share with you links to the prompts and responses, then we will do a brief summary of the responses and logic. All of these were conducted on March 7, 2026.
Most Popular Stories
Most Popular
I encourage those doing significant explorations using AI for research and related purposes to try out the same prompt on each of the three leading AI providers as above. This helps to give a better-rounded response, tapping the strengths of each as well as making any hallucinations, biases or other unwanted artifacts more obvious. Those who are particularly interested in this topic should visit each of the URLs listed above to get a comprehensive view of what we should expect in higher education employment in the next four to five years.
OpenAI’s latest version responded first with the worthwhile observation:
“The most vulnerable positions are not the most prestigious ones; they are the most routine, rules-based, text-heavy and workflow-driven ones. In higher education over the next five years, agentic AI is more likely to reconfigure jobs and shrink headcount in selected functions than to eliminate whole professions at once. That is consistent with broader labor-market evidence and with recent higher education surveys: the World Economic Forum projects substantial job disruption by 2030, while EDUCAUSE finds that institutions already see AI’s biggest opportunities in automating repetitive processes, offloading administrative burdens and analyzing large datasets.”
These points are well taken. We are most likely to see a reconfiguration of positions with AI augmenting significant portions of full-time positions. This will result in fewer but more productive employees, with many heavily using AI.
Anthology’s Sonnet 4.6 notes, “This is one of the most consequential career planning questions in higher education right now.” It goes on to caution, “The blunt truth: Do not pursue a career in higher education administration in a transactional or process-heavy function. The hiring that sustained those roles for decades will contract sharply in the next 5 to 7 years and entry-level pipelines will thin first.”
Sonnet is very honest in its replies, painting a difficult picture for those who seek to find new jobs in higher ed. For those already in the field, Sonnet suggests becoming the most adept user of AI in your office. Seek to transfer to the unit or office where AI is a top priority. It adds, “Consider whether your institution is viable. Smaller, tuition-dependent institutions without strong endowments are in structural decline. Loyalty to a sinking ship is not a career strategy.”
Gemini 3 Thinking Mode responded to my prompt,
“As we navigate this ‘Agentic Era’ of higher education, a transition you’ve been closely monitoring at UPCEA, the landscape for university careers is shifting from the management of tasks to the management of outcomes. The emergence of agentic AI (systems that don’t just generate text, but autonomously execute multi-step workflows) is fundamentally redefining institutional efficiency. By the end of this decade (2030), we will see a ‘silicon-based workforce’ handling much of the operational churn.”
Gemini 3 goes on to say,
“The most vulnerable roles are those defined by structured, repetitive and rules-based workflows. Agentic AI excels at ‘nurturing funnels’ where a student or process moves from Point A to Point B without needing high-level strategic intervention.”
Looking at the remainder of the decade, Gemini suggests that “such positions as Admissions Document Verifiers and Recruiters; Financial Aid Processors and Compliance Officers; and in Academic Support TAs (Grading/Q&A), Librarians (Cataloging) are all vulnerable.”
Across all career stages in higher education, Gemini recommends, “To remain relevant, higher education professionals must pivot toward AI Orchestration. Success is no longer measured by how well you perform a task, but by how well you direct the agents performing them.”
This reinforces the common wisdom that programming is not the essential skill that will dominate the workforce, rather, it is how to orchestrate and manage the available AI tools to achieve useful and efficient outcomes. It is likely that both formal and informal continuous professional development will be required to succeed in the emerging environment in higher ed.
Sonnet 4.6 Extended summarizes,
“Your value proposition must shift from ‘doing’ to ‘judging, contextualizing and leading.’ Mid-career professionals who have not moved from execution to oversight are most at risk. If you’re still doing work that looks like entry-level work plus experience, that’s a problem. Position yourself as an AI implementation leader. Who at your institution will evaluate vendor AI tools? Manage the transition? Train staff? Handle the ethical and equity questions? That person needs your institutional context and trust. Make it you. Pursue leadership development intentionally. If you haven’t already moved into management, do it now — not because management is immune, but because leadership requires the organizational trust that AI systems cannot earn.”
ChatGPT 5.4 Thinking Mode summarizes a key to assessing sustainability among jobs: “A more precise rule of thumb: A university position becomes more vulnerable as more of its value can be described as: ‘Receive information → summarize it → apply rules → generate standard output → route it onward.’” However, “A position becomes less vulnerable as more of its value can be described as: ‘Diagnose ambiguity → exercise judgment → build trust → persuade humans → handle exceptions → take responsibility for outcomes.’” That cuts across all positions in higher ed and many other fields.
Specifically regarding faculty, ChatGPT 5.4 says,
“Faculty whose work is heavily lecture-delivery, generic content generation, low-authenticity assessment and standardized feedback are exposed. Faculty whose work centers on coaching, seminar leadership, studio critique, clinical supervision, research judgment, community-engaged learning and complex assessment become more valuable. EDUCAUSE found that faculty were especially active in using AI for creating learning activities and assessments, which indicates that instructional work itself is already being redesigned. The vulnerable part of faculty work is not expertise; it is the repeatable packaging of expertise.”
Gemini 3 Thinking Mode closes with the sharing of a bit of transformative wisdom: “The ‘Co-Bot’ Workforce Formula: In this new economy, professional value ‘V’ is no longer a function of effort ‘E’, but a function of human judgment ‘J’ applied to agentic output ‘O’ Thus: V = J * O.
If J (judgment) is zero, the value is zero, no matter how great the output.
