At This D.C. Charter, AI Permission Changes With the Assignment

Washington Leadership Academy has moved generative AI beyond a classroom experiment. The D.C. public charter uses it in lessons, student support, teacher work, and attendance operations. In a February 2025 survey, 85 percent of teachers said they used AI professionally, and more than 80 percent said they used it daily or weekly. The school’s distinguishing move is to make student use contingent on the assignment, rather than treat AI access as an all-or-nothing choice.

Permission is not the same as a policy

The school’s approach begins with exposure and boundaries. All freshmen take a computer science course on core AI and machine-learning concepts. Teachers then use a rubric that sets permitted AI use for each assignment on a scale from zero, for no use, to four, for unlimited use. That makes AI access a decision tied to the task rather than a standing permission.

WLA began exploring the technology shortly after ChatGPT’s November 2022 public release. An AI task force of administrators, students, and teachers tested tools and use cases, while staff members led professional-development sessions during the 2024–25 school year. The single-site charter’s structure and its purchase of premium AI tools for teachers helped the experimentation move faster.

Students are also being asked to build with the technology. About 100 took part in a hackathon centered on AI chatbots for real-world needs, producing projects that included a student-government bot for conveying concerns to school leadership.

The lesson plan puts the bot inside the exercise

Several classroom uses make a chatbot part of the exercise, not its final authority. In an AP Government simulation, students designed fictional countries and used a bot to test whether they could resist conquest; the tool gave an outcome without explaining it, leaving students to reason through the result. A math-practice chatbot supplies hints and feedback, but it sometimes gives wrong answers, requiring students using it to defend their reasoning against the system.

The feedback model extends to AP Psychology. Teacher Adam Browning built a tool to generate practice questions modeled on the limited questions released after the College Board overhauled the course curriculum. Students receive immediate feedback, and the chatbot flags Browning when they keep struggling; the school says practice-question scores have more than doubled. That is a classroom performance claim, not a demonstrated long-term learning result.

The research summary cited in the school’s case study points in the same direction. A March 2026 Stanford review of 14 studies found mixed results: students often performed better while they could use AI, but gains often disappeared once access was removed. The review found more promise in tools with guardrails, such as tutors that offer hints rather than answers, while warning that AI can ease academic work at the expense of deeper thinking and retention.

AI reaches the school office, too

Teachers also use chatbots for parent newsletters and other administrative work. On the operations side, WLA centralizes daily attendance data with historical absence records in an AI-supported dashboard, allowing staff to identify and contact students before truancy becomes chronic.

The expansion does not settle the school’s underlying tensions. Staff and students continue to raise concerns about cheating, inaccurate outputs, job security, creativity, and the loss of human connection. WLA’s model is built around constrained use and classroom tools that can challenge students’ thinking; whether it delivers durable learning remains unmeasured.