Elon University teaches its analytics students to distrust AI: framing the problem well is the skill that counts

By Elon University · November 2, 2026.

Elon University's Martha and Spencer Love School of Business (North Carolina) held its Business Analytics Summit on October 29, under the theme “Future of analytics: driving career success through responsible AI and analytics.” It brought together students, faculty and industry professionals for sessions, a data challenge, a poster competition and an industry panel, according to the university's own account.

In the session on responsible AI, alumnus Peyton Pesavento, an enterprise AI strategy consultant at Accenture, urged students to understand the problem before asking an AI for help. “Frame the real problem,” he said, advising against simply telling the model “I'm trying to do this, help me.” He and his colleague Jared Skinner recommended verifying the information given to the AI, questioning its answers and bringing real work samples to interviews. Beca Mayr, vice president of talent acquisition at Sedgwick, explained how she combined her human resources experience with technical skills to build AI agents that support recruiting.

In the data challenge, teams analyzed public housing and urban development data from Alamance County (building permits, tax parcels, municipal boundaries and other sources). The winning team, made up of Takoda Moore, Sebastian Zucker-Malone and Shayne Goldstein, first tried to predict where development would occur. When their models failed to beat a simple baseline, according to the university, they changed the question and studied where it was least likely. Another award-winning project in the poster session examined the use of AI in hospitals.

The afternoon panel included leaders from Franzy, Wells Fargo, The Link Group, the City of Charlotte and Lenovo. April Frazer, of Wells Fargo, said AI could save her team time gathering financial data so it can focus on interpreting the numbers, and that she had to correct a tool that gave wrong information about the bank's results. “You have to be testing it and start questioning it,” she said. Elin Walton, of Franzy, summed it up: “It's the humans using the technology who create the value,” which in turn places a different responsibility on the user: providing the right context and data. Brandon Richard, of The Link Group, spoke about helping new employees use AI while still developing critical thinking and client relationships.

Our take: it is a small story, but it captures well where training is heading. When AI does the calculation, the junior analyst's edge is no longer execution, but framing the right question and spotting when the answer doesn't hold up. The winning team's anecdote is the best example: a model that doesn't beat a trivial baseline is not a failure, it is a signal to rethink the problem. That skeptical discipline is exactly what no tool provides on its own.

Overall, the short-term risk to entry-level analytics jobs is real: if AI absorbs data gathering and first-pass analysis, traditional apprenticeship roles narrow, hence the speakers' emphasis on portfolios of work samples and independent judgment. That a university is already addressing this in the classroom is a good sign. In the long term, freeing analysts from mechanical tasks so they can focus on interpreting and deciding points exactly in the direction we advocate: less repetitive work, more work that requires judgment. It is worth not overstating the scope, however: this is an institutional account of an event, with no employment outcome data yet showing that this approach works.