Schools & teaching
For Jen Roberts, AI means another chance to give useful feedback.
A San Diego English teacher describes using AI to support revision while keeping teaching and judgment in human hands.
For an English teacher, a class full of drafts is also a class full of opportunities to help. The challenge is reaching each student while the work is still fresh.
The need: feedback while there is still time to use it
Feedback has a practical purpose: help a learner decide what to do next. A comment that identifies a weak claim, asks for evidence or points out an unclear transition gives a student a route into revision. A score alone tells them much less about how to improve.
The challenge is giving that attention across a class of different drafts and different needs. Some students need help getting started, others need to develop an argument, and others need a reader who will challenge their assumptions. Teachers also need to distinguish a misunderstanding of the subject from difficulty expressing an idea.
The role of AI in feedback is practical: help teachers identify where students need attention and give learners specific questions to work through during revision.
A teacher's account from San Diego
In a February 17, 2026 Scientific American interview, Point Loma High School teacher Jen Roberts described using AI as another reader of student work. She compares its suggested scores with her own and uses education-specific tools configured with an assignment and rubric to help students revise. She reports returning writing sooner and enabling more feedback within a lesson. These are an experienced teacher's observations, not a controlled study of achievement or grading fairness. Read the interview.
Roberts’s account identifies a clear need: timely, thoughtful feedback that teachers can review without adding another layer of administrative work.
Support preparation, drafting and review
Before an assignment, a teacher would use AI to prepare examples of a clear claim, an unsupported claim and a paragraph with evidence that does not quite fit. The class would discuss the differences. The teacher would check the examples against the lesson and decide which ones to use.
During drafting, an approved tool would ask a student to locate the evidence for a particular sentence or explain how two paragraphs connect. The student would revise, compare the versions and describe what changed. That preserves an opportunity to think instead of turning every comment into a replacement paragraph.
After a round of review, a teacher would examine recurring questions across work they are authorized to process. If several students confuse a summary with an argument, the teacher would respond with a short class lesson rather than more individual generated comments. The teacher would verify that the apparent pattern is real before changing instruction.
Feedback needs to respect the learner’s voice
A useful reviewer must recognize that polished language and sound reasoning are different qualities. It should also avoid flattening every student’s style into the same voice. Learners need room to disagree with a suggestion and explain why they kept their original wording.
The teacher would remain responsible for assessment. Any school application would need clear handling of student drafts, access for the appropriate adults and a way to correct inaccurate feedback. Students should understand what the tool is doing and what information they are providing.
A practical first test
An program-supported pilot should compare the quality and timeliness of feedback, teacher workload and students' later independent writing. Educators should record when AI comments were wrong or unhelpful. Students' drafts would need an approved data-handling process rather than being placed into an unrestricted public chat.
Roberts's account gives educators a concrete approach to evaluate. American Super Intelligence has not established a classroom service or a relationship with her school.
How Liberty will give educators more control over the process
LibertyGPT will give teachers practical tools for preparing examples and reviewing suggested feedback. Liberty OSS will let qualified school and research teams examine model behavior, adapt it to approved materials and test it against the questions that matter in their classrooms.
The OSS pathway will make that work reusable: an evaluation that exposes poor feedback on evidence or reasoning can help teams improve the next release. A school will still need an approved application and privacy arrangements; a public chat alone will not establish those protections.
the program’s education commitment will support both teachers and students. Lessons will teach educators to evaluate the assistance and give students practice in explaining their revisions. Liberty will help by making tools, model access and instruction available together, with the educational goal kept in view: better independent writing.
Explore high school and professional learning for educators.