What happened when History 2C added Kai Coach.

Same course. Same instructor. Same prompts.

History 2C at UCLA is a six-week survey of medieval and early modern European religious history taught by Professor Teofilo F. Ruiz. The course ran in Summer 2025 without Kai Coach and in Summer 2026 with it, under the same instructor, the same six-week structure, and prompts that matched word for word. In 2026, Kai Coach supported four modules across the term: the weekly discussion posts, the first paper, the peer responses, and the final exam. We analyzed every written component from both years, 980 pieces of writing in all. We did not use grades, because the grading was not blind to the coach; every measure comes from the submitted text itself.

The gain reached both ends of the class.

Papers with no developed argument fell by more than half, from 34.7 percent to 15.8 percent. At the same time, papers with four or more developed arguments roughly tripled. The spread of the class barely changed, which means the whole class shifted upward rather than the strongest students pulling away. On the final exam, all five questions that appeared in both years favored the coached cohort, across 143 student answers. Kai Coach did not simply make strong students stronger; it raised the work of the entire cohort.

The writing looks less machine-made, not more.

The obvious worry about putting an AI tool into a writing course is that the essays will come out reading as though a machine wrote them. The opposite happened. Machine-typical vocabulary fell from 0.46 to 0.18 instances per thousand words, a drop of more than 60 percent, and em-dash use fell from 2.07 to 0.93 per thousand words. In an anonymous survey, four of five respondents said they would otherwise have used a general-purpose AI tool, yet the coached cohort's essays read as less machine-generated than those of the cohort without a coach. These markers offer evidence against ghostwriting rather than proof of its absence, and we report them that way.

Why did witch hunts rise as heresy trials declined?

A truism, a cliché, and a definition appear before the answer touches the question.

A causal mechanism, located in institutional authority, appears in the first sentence.

Every coach turn asked a question.

Across 389 instrumented discussion and peer-response sessions in the History pilot, every single coach turn contained a question. Kai Coach never wrote a thesis, a sentence, or a paragraph for a student. When students stalled, it supplied a term or a definition and then handed the reasoning back to them. This record matters because it shows that the coach held its rule under real classroom pressure, not only in a demonstration. The quality of each exchange, not the number of passes, drove the improvement, so students may now use Kai Coach as often as they wish.

A two-cohort comparison, reported with care.

This study compares two cohorts of one course; it is not a randomized trial, and the cohorts differ in year and in size. Of the content measures, the rise in mechanism reasoning on the final exam held up after correction for multiple comparisons. The other figures report observed differences between the two years. New studies will test Kai Coach in Physics and in Comparative Literature, beginning in Fall 2026 and Winter 2027. Instructors and administrators may request the full pilot report from Kudu.