As higher education navigates the AI disruption, concerns about academic integrity and the fate of homework often dominate headlines, but I think a more significant threat might be the erosion of the first rung on the career ladder for our recent graduates. 

Every spring, Minnesota's colleges and universities send thousands of graduates into a labor market that is shifting under their feet, and the educators and workforce professionals who support that transition have a shared stake in understanding what is changing and why. We dedicate four years to preparing students for that foundational first step, so they have a stable foothold as they enter professional life. 

The early stages of almost all professional careers have been defined for decades by routine, entry-level roles where our recent graduates learn the fundamentals of their field by doing the "grunt work" and working closely with more senior professionals. AI can take on some routine tasks, reducing the number of entry-level workers who are needed. Ethan Mollick describes this disruption as an undermining of the hidden system of apprenticeship that comes after formal education. Generative AI threatens this essential apprenticeship layer by changing and elevating baseline expectations, creating a need for higher education to reconsider how we equip students to thrive in dynamic workplace environments that value new skill sets and approaches to work. 

A Tough Job Market for New Grads

The Federal Reserve Bank of New York reported recent graduate unemployment at 5.7% in the fourth quarter of 2025 and underemployment climbing to 42.5%, its highest level since 2020. The National Association of Colleges and Employers found that nearly half of employers rate the job market for the Class of 2026 as “poor” or “fair,” the most pessimistic outlook since the first year of the pandemic. Minnesota’s unemployment rate increased from February 2025 to February 2026 by one percentage point from 3.5% to 4.5%. A breakout for ages 22-25 isn’t available at the state level.

While different studies come to different conclusions about the impact of AI on employment and a variety of economic factors are at play in the current job market for recent college graduates, the college class of 2026 is entering a more challenging hiring environment than those who graduated in the years before them.

AI Job Displacement and Job Augmentation

There are growing concerns that AI is displacing entry-level jobs typically held by recent graduates, narrowing the first rung of professional career ladders and creating a more competitive employment landscape. A Stanford University study published in November 2025 by economist Erik Brynjolfsson and colleagues offered the first large-scale empirical confirmation of these concerns. Drawing on payroll records from millions of workers, the researchers found a 16% relative decline in employment for workers ages 22 to 25 in the most AI-exposed occupations since late 2022, even as employment for older workers in the same fields grew or held steady.

Others have raised concerns about AI’s impact on positions with high exposure to AI. In March 2026, BlackRock CEO Larry Fink publicly warned that the Class of 2026 “could experience the highest jobless rate in years” due in part to AI. Routine tasks such as basic data entry, document reviews, and customer support interactions, historically foundational to entry-level positions, are rapidly being automated. SignalFire reported that hiring of new graduates at the largest tech firms has fallen more than 50% since 2019, with new graduates now making up just 7% of Big Tech hires, down from 15% before the pandemic. It is estimated that up to 66% of entry-level finance jobs could be eliminated due to AI's ability to handle numerical analysis, report generation, and computer coding tasks. Wall Street banks, including Goldman Sachs through initiatives like "OneGS 3.0," are actively exploring AI-driven replacements for junior analyst roles. Similarly, administrative roles in health care face substantial exposure to automation, with projections that up to 80% of such administrative tasks could be automated by 2029.

AI is also augmenting entry-level roles in significant ways by empowering human workers to perform more complex and higher-value tasks, enhancing efficiency, and broadening their capabilities. New hires must immediately engage in more strategic, creative, and people-oriented tasks alongside AI. For example, KPMG now assigns recent graduates complex tax work previously reserved for employees with two to three years of experience. Anthropic’s March Economic Index report found that in February 2026, 49% of jobs in eight key occupational groups could use AI to complete at least a quarter of their tasks, up from 36% in January 2025. The Anthropic report also noted that usage on its API platform is leaning increasingly toward automated workflows rather than human-led collaboration. In healthcare, AI is transforming entry-level roles by automating routine administrative tasks, freeing these professionals to do more direct patient care while AI-powered chatbots and virtual assistants can now handle initial triage, gathering basic health information, and remote consultations.In education, AI tools automate routine tasks such as grading and initial lesson planning, enabling educators to concentrate on direct student engagement and higher-order teaching responsibilities. 

AI's displacement and augmentation of entry-level roles create heightened expectations for new graduates as employers across industry sectors place greater value on AI literacy, critical thinking, and essential soft skills. Universities should be rethinking educational approaches to prepare students for these elevated workplace requirements.

Recommendations for Rethinking the First Rung of Professional Career Ladders

I’m asserting that colleges and universities can no longer rely on incremental tweaks to prepare students for an AI-driven workforce.  We must balance academic freedom with our academic responsibility to redesign curricula along with teaching and assessment practices to ensure all students are AI literate and possess domain-specific technical fluency, and distinctly human capacities.

Curriculum redesign 

• Foundational AI Literacy - We should build agreement among faculty groups about what AI fluency means within the context of our respective missions and student populations. Once this baseline fluency is clear, we can embed it across all majors, whether through a required introductory course, scaffolded modules in general education classes, or micro-credentials that certify core competencies. Beyond this foundation, academic departments should weave domain-specific AI applications into existing courses. For instance, finance students might learn Python scripts for automated data pipelines and marketing communication students might learn AI-powered, automated A-B testing for micro-targeted copy. 
• Flexible and Relevant Degrees - Through 25 years of working in academic settings across a variety of public and private institutions, I know the transformative value of higher education and the slow pace of change to update curriculum and launch new programs. We need to be more nimble and accelerate academic processes, so that degree programs are reviewed more often and aligned with quickly evolving industry expectations. We should design more interdisciplinary pathways such as "Finance + AI," "Marketing + AI," or "Nursing + AI" majors or minors that allow students to blend subject matter expertise with technical know-how. Curricular pathways should be reviewed for opportunities to integrate stackable micro-credentials and industry certifications, so graduates have industry-recognized evidence of skill. 
• Real-World Learning Experiences - Throughout my career, I’ve often heard the phrase “in the real world” which implies that the experiences we’re providing students are not reflective of what they will encounter after graduation. We should ask ourselves why this phrase reoccurs so often and the message it sends to increasingly savvy students. All students should have opportunities to engage in meaningful experiential learning, including internships, co-op placements, undergraduate research opportunities, and project-based capstone experiences. Every course should help students integrate classroom knowledge with real-world practice. 

Pedagogical shifts

• Instructional Efficiency - Based on Spring 2025 survey data, we know that 58% of our faculty are using AI at least monthly to lighten their instructional workloads by helping them draft assignments, lecture notes, slides, quizzes, exams, syllabi, and project timelines. Auto-graded quizzes and chatbot tutors can handle low-stakes assessments and routine student inquiries, freeing faculty to concentrate on deeper feedback and mentoring that focuses on higher-order skills such as critical thinking and ethical reasoning.
• Personalizing Learning - We’re helping faculty personalize student learning by training them to build custom AI-powered agents and tools that embed course content and reflect their pedagogical style. Using the Mollick and Mollick blueprint prompts, an instructor can spin up an AI tutor or teaching assistant that quizzes students at the right level, offers instant formative feedback, and surfaces misconceptions the instructor could address, all without writing code. Because these tools work across GPT 5-class models such as Gemini and Claude, faculty can choose the platform their campus supports. The same prompt engineering approach lets them create hyper focused micro games such as a negotiation simulator, a goal setting role play, or a data analysis escape room that respond to each learner’s moves, keeps engagement high, and captures rich formative data. Since the instructor owns the prompt, they can adjust persona, difficulty, and feedback rules on the fly, ensuring the chatbot reflects their standards while freeing class time for higher order discussion and mentoring.
• AI in Assignments - Coursework could weave AI directly into learning tasks by asking students to use AI tools to brainstorm, draft outlines, and interrogate AI-generated outputs by combining practical tool use with careful human evaluation. At the same time, we can emphasize active, problem-based learning through case studies, simulations, and open-ended projects that allow students to tackle real-world constraints while building creativity, collaboration, and critical thinking. Integrating generative AI into these experiences can deepen learning by helping students explore design options, simulate stakeholder input, and analyze data. These tools enhance problem-solving while reinforcing the ethical reasoning and human judgment that AI cannot replace.
• Assessing Learning - Because generative AI can replicate the typical academic work used to evaluate student learning, we need to design assessment practices that focus on authentic, performance-based tasks where students must apply knowledge in real-time, defend their choices, and provide a transparent record of their learning process. When AI tools are used, students should record each step, cite AI-generated contributions, and reflect on how, when, and why AI was used. We should expect more from our students given the AI tools they have at their disposal; however, we need to rethink how we evaluate student learning, so these powerful AI tools do not undermine the learning process. 

Connecting industry expectations with dedicated investments in faculty development can help us rethink the first rung of the career ladder so we intentionally graduate students who possess the right mix of AI competencies with uniquely human skills grounded in their discipline. 

Of course, all colleges and universities are wrestling with how to leverage AI in learning and how to prepare students for a world impacted in countless ways big and small by AI.

The University of St. Thomas is making strong progress on this front through our Institute for AI for the Common Good which provides a mission-based framework for coordinating AI-related activities across the university by evaluating ongoing initiatives, overseeing research and development projects, and identifying new opportunities for innovation and collaboration. The Center for Strategic Transformation in Education, Learning, And Research (STELAR) provides comprehensive faculty training and pedagogical resources (e.g., hands-on workshops, faculty learning communities, badges, intranet resource hubs) for effectively integrating AI across curricula. These initiatives are reinforced by new academic programs focused on harnessing AI's potential and over40 course offerings this academic year focused on AI. 

By putting these changes into practice, we give students a surer foothold on that crucial first rung of the career ladder, elevating the value of their degrees and positioning higher education as an engine for producing adaptable, ethical, AI-fluent professionals who can thrive amid accelerating technological change.

Jonathan Keiser, PhD
Associate Vice President, Academic Technology, AI Enablement and Innovation
University of St. Thomas (MN)
keis2759@stthomas.edu
 

Author's Note: This article was developed through an iterative research and writing process that used generative AI tools to gather, interrogate, and synthesize information from a wide range of sources. Curated research materials and web sources were organized in NotebookLM, where its interactive features helped me pressure-test claims, refine the argument, and identify gaps. Readers are invited to explore the supporting materials in NotebookLM to draw their own connections and insights.