Introduction
In all circumstances, the teacher-student relationship will continue to be central to students’ experience of school.
We believe that student-facing AI-powered products should be considered in light of two questions:
How does the student learning experience with this product compare to the learning experience that would otherwise typically be happening (i.e., teacher-delivered instruction without AI)?
How does the student learning experience with this product compare to the very best learning experience we wish all students had access to in this discipline?
Both questions matter. We want to see better learning outcomes and faster growth rates than we are currently. We especially want to see the students who have the least access to consistently strong instruction get access to better learning experiences. If AI-powered tools can help students learn more, we believe we should leverage them.
The promise of any instructional tool must be anchored to giving students the very best possible learning experiences.
That is, the bar shouldn’t simply be “are these learning experiences better than the current average?”; it should be “are these learning experiences consistently excellent and delivering on the best of the human experience?” Every product that students use directly should demonstrate a positive impact on learning overall and for every student group. Students who have historically been held to the lowest standards benefit the most when instruction raises the bar and supports them in meeting high expectations.
Products should strive for disproportionately positive impact for students of color, students experiencing poverty, multilingual learners, and students with disabilities and that product developers should publicly report their impact data by student group.
Every product that students use directly should demonstrate a positive impact on learning overall and for every student group. Students who have historically been held to the lowest standards benefit the most when instruction raises the bar and supports them in meeting high expectations.
We hope this analysis gives a practical view into the benefits and risks AI-powered products pose for instruction. Above all, we hope it inspires instructional leaders and educators to dig in and learn more. The next few years will be fascinating and influential as we see new innovations and develop our understanding of what products can really do for student learning. We look forward to continuing to learn, together.