Sown To Grow's Use of AI and Student Data
At Sown To Grow, we share a deep commitment to protecting student privacy and believe that data protection must be a top priority when evaluating tools used with students. Below is an overview of our practices and philosophy in these areas.
Student Data Practices
Sown To Grow is a check-in, and reflection platform designed to ensure every student feels seen, supported, and able to thrive. The platform has been funded and validated through federal research grants from the U.S. Department of Education, the National Science Foundation, and nonprofit partners.
What data we collect:
- Weekly emotional check-ins (emojis) and short reflections in response to prompts.
- Educator feedback on student reflections.
- No demographic or personal information (race, IEP status, income, etc.) is collected.
How data is used:
- Support student well-being and identify trends.
- Enable educators, counselors, and administrators to provide timely, personalized support.
- Inform schoolwide efforts to strengthen learning environments.
Key protections are in place:
- Only the student, their teacher, and approved administrators have access to data.
- No demographic or personal information (race, IEP status, income, etc.) is collected.
- Data is never sold or used for advertising, nor shared with third parties unless explicitly authorized by the district.
- All practices align with COPPA, FERPA, the California Student Privacy Alliance, and the National Student Data Privacy Consortium.
- District contracts and data-sharing agreements always supersede general policies.
- Student data is stored securely in the U.S.-based cloud systems (AWS, Databricks) under strict compliance standards (SOC 2, ISO 27001, FISMA, etc.).
- Lastly, we undergo regular penetration and security testing to stay up to date.
Use of AI
Sown To Grow does not use AI to interact directly with students. Instead, AI is used to support educators by surfacing insights and providing optional tools such as suggested responses designed to help when time is limited. Educators retain full control at every step.
Key AI features:
- Concerning Reflection Alerts – Student reflections with concerning language or elevated stress are flagged for educators after human review by trained staff.
- Feedback Suggestions – Teachers may see optional, customizable response suggestions to student reflections (never shared directly with students).
- Reflection Quality Assessments – Reflection depth and quality are assessed against educator-developed rubrics to guide coaching opportunities.
Our AI models:
- Built fully in-house using traditional, interpretable machine learning models (Random Forest, XGBoost, Isolation Trees, KNN).
- Trained only on anonymized, de-identified reflections and aggregated educator feedback—never on individual student data or district-specific records.
- They are designed to prevent any possibility of tracking or long-term data storage.
- Continuously monitored with accuracy, precision, recall, and F1 score; low-confidence outputs are manually reviewed and incorporated into retraining cycles.
- Designed for transparency, fairness, and educator oversight, not generative outputs.
- No generative AI systems are in use today. If introduced in the future, districts and educators will have clear control to enable or disable them.
Our ethos around AI:
While AI can unlock tremendous value, using it to make recommendations, and support decision-making brings with it the chance of unintended consequences. STG strongly believes that a student’s education is too important to take these concerns lightly. To ensure that they are minimizing the chance of unintentionally causing harm, STG has a strong ethos regarding their use of AI:
- Educator Oversight: STG does not display any components or outputs of its AI technology directly to students; they always keep an educator involved in the process.
- Transparency and Trust: Outside of trade secrets, STG seeks to share the logic behind decision making (ex: reflection quality rubric, alerts logic) with educators. They also ensure there is an open line of communication regarding any questions or concerns.
- Self Scrutiny: STG applies a high level of rigor when developing AI tools, questioning potential implications at every stage - from model features to product design. This principle underpins the following practices:
- Strong Research & Pedagogy Phase: AI features are co-developed with educators, researchers, and mental health experts to ensure alignment with sound pedagogy and research.
- Build Incrementally & Collect Feedback: Tools are rolled out in stages, with ongoing educator feedback used to validate, refine, and improve the approach.
- Performance vs. Explainability: When faced with a tradeoff, STG prioritizes explainability over maximum accuracy, ensuring educators understand how outputs are generated.
- Bias Awareness: Models are closely examined for potential biases in their training data, with proactive steps taken to mitigate inequities.
- Team Diversity: A diverse team helps uncover blind spots and brings multiple perspectives to decision making. All members of the STG team, regardless of how technical their roles / backgrounds are, share this ethos for AI, and the responsibility for upholding it.
Environmental Responsibility
We are mindful of technology’s environmental impact. Our AI models are lightweight, efficient, and developed in-house to minimize energy use. We avoid large, general-purpose models that consume excessive resources.
Independent Privacy Ratings
We value independent evaluations like those from Common Sense Media. In 2023, their Privacy Program awarded Sown To Grow a blue “Pass” rating - one of their highest. We are actively working with their team to ensure public listings accurately reflect this status.
Design Philosophy
Sown To Grow was built by educators, researchers, and technology experts with a shared mission:
- To build tools that genuinely serve students and educators.
- To protect student well-being and privacy above all else.
- To prioritize sustainability and avoid unnecessary technology for its own sake.
