In a review of 126 K-12 ML activities, supervised learning activities cluster overwhelmingly in the lowest algorithmic-abstraction levels¶

Claim · All claims
Evidence · 1 study · theoretical r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? A systematic review of 126 K-12 ML activities coded by levels of algorithmic abstraction found invisible use, button interaction, and model deployment account for almost 60% of coded supervised learning entries, with open-ended creation tasks appearing only twice. → Pierre Haritz 2026

Evidence¶

Pierre Haritz 2026¶

Pierre Haritz, Hendrik Krone, and Thomas Liebig. (2026). Addressing Trust in AI Systems through Education: A Didactic Perspective. https://arxiv.org/abs/2609.02453 doi:10.48550/arxiv.2609.02453

q2 · i? · theoretical · r2

Systematic review of 126 K-12 ML activities coded against a taxonomy of algorithmic abstraction ranging from invisible use to independent creation. The review reports the lowest abstraction levels account for "almost 60% of all coded supervised learning entries"; no inferential statistics are printed.

"invisible use, button interaction, and model deployment together account for almost 60% of all coded supervised learning entries, while genuinely open-ended creation tasks appear only twice in the entire corpus"

Discussion¶

Related Claims¶

• Across 133 K-12 ML activities, a technical and a societal perspective on the same algorithm co-occur in roughly 3% of cases — related
• Of 89 coded supervised learning activities, 69 confine learners to the User role, 17 reach Creator, and 8 reach Machine — related