Calabi Labs · Guide · 2026-06-23
Michigan ross research team creates ai detection tool michigan ross
Michigan Ross Research Team Creates AI Detection Tool Called Liketropy
A research team at the University of Michigan's Stephen M. Ross School of Business has developed Liketropy, an AI detection tool that identifies AI-generated text with over 96% accuracy—while maintaining a remarkably low false accusation rate against human writers.
Doctoral candidates Tara Radvand and Mojtaba Abdolmaleki led the project alongside collaborators, creating a tool specifically designed to solve one of the most persistent problems in academic and professional settings: reliably distinguishing AI-written content from human writing.
Why Existing Tools Fall Short
Most AI detectors struggle with two critical failures modes: false positives (flagging human work as AI-generated) and false negatives (missing AI-generated content). A false accusation can unfairly damage a student's grade or an employee's credibility. Existing tools often produce inconsistent results across different AI models and writing styles.
Liketropy addresses these shortcomings head-on.
How Liketropy Works
Unlike traditional AI detection systems that require extensive training on labeled datasets, Liketropy achieves high accuracy without prior training data. The tool examines text patterns and linguistic fingerprints that emerge differently in AI-generated versus human-authored content—even when the AI output has been edited, paraphrased, or slightly modified.
The system performs best when evaluators have a sense of which AI models might have generated the suspicious text. Under those conditions, it delivers that 96%+ accuracy with a false accusation rate low enough for practical use in grading, hiring, and editorial review.
What This Means for Educators and Employers
The implications are significant. Instructors can use Liketropy to evaluate student submissions more fairly, reducing the guesswork and bias that come with gut-instinct detection. HR professionals and editors gain a defensible, data-driven layer of scrutiny.
The tool doesn't eliminate judgment—it augments it with objective measurement.
The Bigger Picture: Responsible AI at Michigan Ross
This research fits within Michigan Ross's broader commitment to responsible AI adoption in business. As generative AI reshapes industries, the school positions itself at the forefront of understanding both the capabilities and the risks these tools introduce.
Liketropy represents a practical response: not banning AI, but building the infrastructure to use it honestly.
Try Calabi free at calabilabs.com — 10 cleans, no card.