Science AI research leaves governance gaps
What does the science-education research base actually cover? A review maps 80 peer-reviewed studies across technology, curriculum integration and ethics. Personalization and adaptive learning dominate, accounting for 47 studies, while privacy and data protection appear as the coded ethical concern in only one. Higher education receives more attention than K-12 or teacher education. These patterns identify gaps in what the selected literature examines; they do not establish that a particular school tool is unsafe or ineffective. The review's proposed framework links the technology being used, its instructional role and its ethical implications. Those relationships are conceptual propositions, rather than demonstrated causes of better learning.
Why it matters for leaders
Editorial recommendation: Ask each science pilot team to complete a short evidence map: the concept students will learn, the activity AI changes, the student information required and the evidence the teacher will collect. Assign curriculum and technology leaders joint responsibility for unresolved questions. An attractive simulation should earn a place through its instructional purpose, with an equally clear account of the information it consumes.
The evidence
The author combined bibliometric and qualitative analysis of English-language studies indexed in Web of Science and Scopus, spanning 1990-2026.
Who was studied
80 research publications; higher education predominates.
What the evidence cannot tell us
Single-author coding and two-database coverage constrain the map. Publication patterns are not effect estimates, and omitted topics may be studied elsewhere.
Read the original sources
Summary and leadership analysis from the Week ending 2026-09-25 brief. The edition’s selection notes explain source access and evidence limitations.