Follow Us on Google
LAGOS — Universities worldwide are losing a battle they cannot win. As artificial intelligence tools like ChatGPT become more sophisticated, the detection software that institutions have spent hundreds of millions of dollars on is failing at alarming rates. False accusations against innocent students are mounting, while savvy students who know how to manipulate the detectors slip through undetected.
Paul Showemimo, a Nigerian entrepreneur and doctoral researcher at University of Potomac in Washington DC, argues that the entire approach is fundamentally flawed. Instead of asking “Did AI write this?” he proposes educators should ask “Does the student understand this?”
We are fighting the wrong war,” Showemimo said in a recent interview. “Detection is an arms race we cannot win. AI will always get better at mimicking human writing faster than detectors can catch up. Meanwhile, we are accusing students who wrote their own work and creating an atmosphere of suspicion rather than learning.”
His Multi Modal Verification Framework, being developed as his dissertation research, represents a paradigm shift from detection to verification. Rather than analyzing text for AI fingerprints, the framework asks students to demonstrate understanding through multiple assessment methods.
The approach builds on a 2023 pilot Showemimo conducted at Hult International Business School in Cambridge, Massachusetts, USA, where he earned his MBA. The pilot, called Edureel, involved 36 MBA students who were asked to explain their written work through short video responses using Socratic questioning.
The results were striking,” Showemimo explained. “Students who genuinely understood their material were confident, articulate, could answer follow up questions without preparation. Students who did not understand struggled immediately. The difference was obvious within the first minute of video.
“The pilot achieved high completion rates with strong student satisfaction. More importantly, it earned formal endorsement from the Dean of Hult’s Boston campus, validating the approach’s effectiveness in an actual academic setting.
“What surprised us most was student reaction,” Showemimo noted. “They told us they preferred this method because it gave them a chance to prove they were smart rather than having to prove they were honest. That shift in framing changes everything.”
The framework he is now developing for his dissertation expands beyond video responses to incorporate multiple verification modalities: adaptive questioning that adjusts based on student responses, real time explanation requirements, consistency checks across different formats, and behavioral analysis that identifies genuine understanding versus memorization or AI generation.”No single method is perfect,” Showemimo said. “But when you combine multiple verification approaches, it becomes exponentially harder to fake understanding. A student who truly learned the material will demonstrate that consistently across all modalities.”
No single method is perfect,” Showemimo said. “But when you combine multiple verification approaches, it becomes exponentially harder to fake understanding. A student who truly learned the material will demonstrate that consistently across all modalities.
“The research comes at a critical time for higher education. According to recent studies, AI detection tools like Turnitin and GPTZero show false positive rates exceeding 50 percent in some cases. Students who wrote original work are being accused, graded down, or brought before academic integrity committees based on unreliable algorithmic assessments.
“The human cost is devastating,” Showemimo observed. “Imagine spending weeks researching and writing a paper, only to be accused of cheating by software. Some students are appealing these decisions for months. Some are giving up and accepting lower grades they do not deserve. The current system is broken
His perspective is informed by extensive entrepreneurial experience building practical solutions to complex problems. Before pursuing his doctorate, Showemimo founded Eklipse Technology, managing over 200 projects across six industries from 2016 to 2022 and generating more than $300,000 in revenue. He also founded Haba Insurtech, raising over $200,000 in seed funding, and Eklipse Consults, which developed the S5 education platform serving 85,000 students across six African countries before its acquisition.
Currently, as President and CEO of Ignite Solutions, he develops AI powered tools for American small businesses, having deployed systems including voice AI and process automation for about 20 clients across insurance, healthcare, construction and religious sectors.
“Every problem I have solved in business required understanding what people actually need versus what they say they need,” Showemimo said. “Universities say they need better AI detectors. What they actually need is a way to verify learning that students will accept as fair and that actually works.”
His academic credentials include his MBA from Hult International Business School and IEEE Senior Member status. He serves as a peer reviewer for top ranked academic journals in artificial intelligence and education, positioning him at the intersection of technology development and educational assessment.
The framework is still under development, with plans to pilot expanded versions at additional institutions. Showemimo is also building technological infrastructure to support implementation at scale, making the system accessible to institutions that lack extensive technical resources.
“The solution has to be practical,” he emphasized. “If it requires massive technology budgets or complete curriculum overhauls, institutions will not adopt it. We are designing for real world constraints, not ideal conditions.”
As artificial intelligence continues disrupting traditional assessment methods, approaches like Showemimo’s that shift from adversarial detection to collaborative verification may represent the future of academic integrity. The question is no longer whether AI will be part of education, but how institutions can adapt assessment methods to verify genuine learning in an AI saturated world.
“AI is not going away,” Showemimo concluded. “We can spend the next decade fighting it and failing, or we can change our approach to focus on what actually matters: whether students learned what we tried to teach them. I know which path makes more sense.”
