The conversation surrounding education and skills is undergoing a significant transformation, particularly in the realm of financial services, as revealed by a recent PwC survey. The study, which gathered insights from over 1,000 U.S. financial executives, highlights that 86% believe training in artificial intelligence (AI) holds more value for new hires than a traditional MBA. Furthermore, 91% of these leaders plan to increase salaries for those proficient in AI, indicating a willingness to offer higher compensation for this in-demand skill. In response to these trends, prestigious business schools like Wharton and MIT are introducing short certificate programs focused on AI leadership, catering to a professional landscape where MBA applications are on the decline and entry-level salaries are shrinking.
For many years, elite education was synonymous with social mobility and career advancement, largely due to the difficulty and expense associated with obtaining such credentials. These degrees were seen as authentic markers of capability, with rigorous admissions processes ensuring only the most capable individuals were accepted, and graduation serving as a testament to refined skills. However, this system is now being challenged from both the employer side, where skills are prioritized over credentials, and from within the educational institutions themselves, where the integrity of these degrees faces scrutiny.
A notable case is taking place at Brown University. Economics Professor Roberto Serrano, after nearly two decades of teaching Welfare Economics and Social Choice Theory, opted for a take-home exam last spring due to the emotional toll on students after a campus shooting. He designed the assessment to promote substantial analysis, yet the average score soared to 96%, with 40 students achieving perfect scores—a sharp deviation from the norm that typically falls between 65 and 80 percent, raising concerns for Serrano and his grading team.
To investigate, Serrano ran the midterm submissions through ChatGPT and discovered troubling similarities between the AI-generated responses and student work, indicating many may have relied on AI tools. Rather than disqualifying the test outright, he created a pathway for validation: if the scores from a subsequent in-person final were consistent with the midterm, both would stand; otherwise, the midterm would be voided.
When the final was administered, 18 students withdrew from the course and 9 others chose not to take the exam, while those who did saw the class average plummet to 48.6%—the lowest recorded in history, with previous finals never below 65%. Nineteen students failed the course altogether, prompting Brown to initiate procedures to investigate potential academic misconduct.
The primary issue at hand goes beyond mere misconduct; it raises fundamental questions about the nature of education in an age dominated by AI. As technology evolves, the definition of independent work requires reevaluation. Past advancements, from calculators to search engines, necessitated changes in educational standards, but generative AI introduces a more profound shift—one that automates comprehensive aspects of cognitive processing, making it difficult to delineate between a student's own insight and that generated by AI.
In the wake of Serrano's findings, the behavior of the students reveals a deeper narrative. Their choices—to withdraw or not take the final—hint at a challenge to the very purpose of traditional assessments. Students may now believe that true competence lies in their ability to work with AI tools, indicating that their understanding of capability has evolved to accommodate contemporary workplace realities.
While it is essential for universities to confront this challenge, it is equally critical to recognize the distinction between misusing AI and relying on it to enhance one’s work. At Brown, the concern centered around misrepresentation—students submitted work derived from AI rather than demonstrating their understanding independently. This issue isn’t confined to the classroom; it mirrors potential ethical dilemmas in professional environments.
Moreover, the disparity in performance between the midterm and final highlights that AI proficiency does not equate to deep understanding. True fluency—with which employers are now equating value—demands the ability to guide AI, analyze its output, and rectify mistakes, all of which require substantial foundational knowledge.
The current educational landscape reflects a precarious balance between an old-world credentialing system, outdated honor codes, and the current job market's emphasis on AI competency. Universities find themselves grappling with how to authentically assess students equipped with AI tools, and many lack the mechanisms or frameworks to do so reliably.
Moving forward, educational institutions must delineate what aspects of learning are expected to remain personal and where AI can be integrated into assessments. Some elements of coursework should focus on unassisted reasoning, while others should embrace AI and evaluate a student’s capability to utilize it effectively. Faculty must also be empowered to report systemic cheating issues cooperatively rather than through burdensome processes.
As Serrano aptly put it, a society can’t afford to let students think that dishonesty is acceptable; doing so undermines societal integrity. Yet, students’ unvoiced sentiments reflect a valid critique of outdated evaluation methods. The institutions that successfully address both viewpoints and adapt their curricula accordingly will likely maintain their relevance, while those that cling to traditional models risk falling behind as the educational landscape continues to shift.
