Why Your AI Strategy Needs Negative Results

Source: Mike Kentz Substack
Author: Mike Kentz
Original source: https://mikekentz.substack.com/p/why-your-ai-strategy-needs-negative

Private backup: the full article text is archived in the private repository at archives/articles/mikekentz-substack-com-why-your-ai-strategy-needs-negative.source.md. It is not published on the public Quartz site.

Summary

Mike Kentz argues that schools and universities should not wait for generic AI research to dictate their AI strategy because institutional context matters too much. Instead, institutions should run controlled, mission-aligned experiments that collect local data about whether AI helps or harms specific students, tasks, and workflows. A key contribution is the argument that negative or null results are strategically valuable: failed AI pilots can prevent bad investments, reveal risks, and lead to better approaches. Kentz illustrates this with his own failed AI tutoring/mentorship idea and with ongoing experiments at the University of Washington, University of Central Florida, and Dayton Public Schools.

Big ideas

Claims

Key evidence and examples

  • Kentz’s failed upperclassman AI tutoring/mentorship pilot revealed the need for more formal structure and risk controls.
  • University of Washington graduate communication students test transcript analysis, peer transcript annotation, and metacognitive AI reflections.
  • University of Central Florida tests AI use in business model creation through student surveys and iterative experiments.
  • Dayton Public Schools’ planned SEL Check-In Bot avoids therapy and functions as a signal flare for counselors using frequency data and dashboards.

Education relevance

High relevance for school and university AI strategy, pilot design, research-practice partnerships, governance, implementation planning, and evaluation culture.

My notes