Sapphire Core Institute
Teachers, against slop
We're a nonprofit writing AI training with teachers, so that nothing reaches a student unless a teacher has read it and would stand behind it.
It takes five minutes and there's no signup. You read six short essays and decide which ones AI wrote.
Try one
An essay on the green light in The Great Gatsby.
The green light at the end of Daisy's dock is one of the most significant symbols in The Great Gatsby. It represents Gatsby's hopes, his longing for Daisy, and his desire to recapture the past. Green traditionally symbolises hope and renewal, but also envy. Positioned just out of reach across the water, the light ultimately embodies the unattainable nature of the American Dream itself.
Did AI write this?
The position
The problem is material nobody authored
By slop we mean material produced in bulk with nobody's judgment behind it. It spreads because it's cheap, and it crowds out better work. Schools had plenty of it before generative AI, from worksheet mills to packets photocopied for a decade. AI made it nearly free.
We don't think banning AI fixes that. We think the fix is authorship: a teacher decides what goes on the page and can explain why each part of it is there.
Why this exists
Schools got the tools before the training
of US public school teachers have received any formal guidance from their school on how to use AI. Six in 10 use it for their work.
Walton Family Foundation and Gallup, survey of 2,069 US public K-12 teachers, February to March 2026.of students say their teachers haven't explicitly taught them how to use AI for schoolwork.
RAND, national surveys of students, parents, teachers, and school and district leaders, published September 2025.US teens say they do all or most of their schoolwork with a chatbot's help. More than half have used one for schoolwork.
Pew Research Center, survey of 1,458 US teens aged 13 to 17, September to October 2025.of activities in 310 AI-generated civics lesson plans asked students to evaluate anything. 45% asked them only to remember information.
Trust, Maloy, Xu and Pelletier, CITE Journal, 2025. Plans generated by ChatGPT, Gemini, and Copilot.The obvious fix doesn't work
of essays by non-native English writers were flagged as AI-generated by at least one of seven widely used detectors. People wrote every one of them.
Liang, Yuksekgonul, Mao, Wu and Zou, Patterns, 2023.of AI-written text was caught by OpenAI's own detector in the company's tests, and it labeled 9% of human writing as AI. OpenAI withdrew it within six months.
OpenAI, January 2023, withdrawn July 2023 for its "low rate of accuracy."Black teens were about twice as likely as White or Latino teens to say a teacher wrongly flagged their schoolwork as AI-generated.
Common Sense Media, survey of 1,045 US teens aged 13 to 18 and their parents, March to April 2024.We don't recommend detectors, and we don't sell them. The exercise tests the alternative, your own judgment, against essays written to get past it.
We're writing the training these numbers say is missing, and we need teachers to help.
I want to helpWhat we are building
Three programs, and where each one stands
All three are at an early stage.
Training
Workshops for teachers, students, and school staff on using AI safely and without producing slop. We're writing them now with volunteer teachers, and none have been delivered yet.
See where the training standsThe anti-slop standard
Questions classroom material should pass before it reaches students. We're writing them now, and teachers will test them on real worksheets.
Why schools need itDemocratic AI
Teachers should have a say in the tools they use, and build some of them, instead of taking whatever a vendor sells. For now it's a principle. We haven't released any tools.
Read the principleBefore you ask
A few common questions
The full page also covers how the training will work and how to get involved.
Read all the questionsAre you telling me to ban AI?
No. Students have it on their phones, so a ban mostly teaches them to hide it, and the honest ones end up penalized. We'd rather teach judgment: when a tool is helping with the work, and when it's doing the thinking the assignment was supposed to build.
Do you recommend AI detection tools?
No. A 2023 Stanford study found that popular detectors flagged essays by non-native English writers as AI-generated far more often than essays by native speakers, and that a reworded prompt got model output past them. Our article on detectors has the details and sources.
Are you against students using AI?
No. A student who has learned to use it well is better off than one who was only told not to. But uses vary a lot. Having a model check your grammar is a long way from having it write the essay.
How will the training be delivered?
We haven't decided. Right now we're planning a mix of short self-paced material and live sessions, and the volunteer teachers will help settle it.
The paperwork
A registered 501(c)(3) nonprofit
You can look us up by EIN in the IRS Tax Exempt Organization Search.
- Status
- 501(c)(3) nonprofit
- EIN
- 41-4847911
- Contact
- contactsapphirecore.org
Get involved
Come help us build this
Whether you teach, lead a school, or just care how AI shows up in classrooms, we'd really like to meet you. Leave your email and one of us will reach out personally to hear what you'd like to do.