Scale Is Not the Same as Educational Value

Large contracts and rapid deployment made headlines this week, but the evidence from students and teachers was much less certain. Buying access for an entire system does not mean the system has agreed on how the tool should support learning.

What Happened This Week

A piece in EdWeek MarketBrief this week walked through Skills for the Future, a pilot now running in classrooms across North Carolina, Indiana, Wisconsin, Nevada, and Rhode Island.¹²

Cal State just renewed its OpenAI deal at $13 million a year, the largest agreement OpenAI has with any university.³⁴

A team led by researchers at Berkeley and Cornell surveyed 95,513 undergraduates across 20 public research universities and published the results in Science last Thursday.⁵⁶

On May 27, Randi Weingarten unveiled a ten-point plan called Devices Down, Eyes Up, Hands-On.⁷⁸

On May 27, Gallup and the Walton Family Foundation published the newest segment of their Teaching for Tomorrow study.⁹¹⁰

California State University recently renewed its ChatGPT Edu contract with OpenAI at roughly $13 million a year for three more years, deploying the tool across 460,000 students and 63,000 faculty and staff.¹¹¹²

What These Stories Have in Common

The largest AI rollouts often measure licenses, logins, and time saved. Those numbers are easy to collect, but they say little about whether students understand more, think independently, or trust the work they are doing.

Koan starts from a simple premise. The final answer is no longer a reliable measure of learning. Schools need to understand how students reached that answer and whether the work made them more capable.

What Schools Can Do

Schools should define educational goals before expanding access. A rollout should measure changes in student reasoning, teacher workload, and learning outcomes, and it should leave room to change course when those measures do not improve.

A technology becomes educational only when its value can be seen in learning.