How this site uses AI

The weekly journal (feedback on your reflections), the Practice Tutor (guiding hints while you think, and a marking summary when you reveal), Explain It, and Ethics Court. Everything else is ordinary server code: quizzes are graded against fixed answer keys, arcade scoring is deterministic, and the map moves on completed work.

It guides rather than answers. The AI features are designed and instructed to respond with the next question or a nudge rather than the finished answer. In Ethics Court it argues the other side on purpose.

It does not set your grades. Quiz results come from fixed answer keys, and points become confirmed only through teacher review. Practice is point-free, so the AI marking there is feedback on your thinking. AI output is never used on its own to decide a school grade or any other significant judgement about a student; that responsibility stays with your teacher.

It is budgeted. Each student has a daily usage limit, shown before you continue, so access stays fair and the service stays sustainable.

AI feedback can be wrong, incomplete, or inconsistent, and it can occasionally slip past its own instructions and say more than it should. Treat it as formative guidance, not as an authority: compare it with the course materials, and ask your teacher when something does not seem right. If a response is incorrect or gives too much away, report it with the in-app feedback tool; those reports genuinely shape how the features are tuned.

EduCS uses Claude models from Anthropic through their API. What gets sent is the task, your submitted text, and the recent conversation needed to generate a response; the app does not send the name you sign in with, your class code, or any account identifier with it. Under Anthropic’s commercial API policy this content is not used to train their models, and Anthropic deletes API inputs and outputs within 30 days, with narrow exceptions for content their safety systems flag and for legal requirements.

Each conversation is also stored in the EduCS database, attached to your account, so you and your teacher can look back at the feedback and so the daily limits work. It stays as long as your account exists and is deleted with it. Teachers can read these conversations, but nobody is watching them live.

The journal and the AI features are learning tools, not a confidential support channel, and nothing you write in them is read by a person the moment you send it. Keep private or identifying information out of them. If something personal or urgent is going on, talk directly to a trusted adult at your school.

The point is to extend the opportunities for feedback between lessons, while judgement, assessment, and the classroom itself stay with the teacher. Questions about any of this are welcome through the contact page on educs.me.