Treat trust calibration as an explicit educational objective of ML education, pursued through ICE-T's three facets¶

Strategy · All strategies
Evidence · no claims cited

Description¶

The article recommends that school curricula promote transparency, data literacy, and an understanding of key algorithmic principles, and proposes that "trust calibration be treated as an explicit educational objective of ML education, alongside traditional objectives such as conceptual understanding and procedural skill". ICE-T supplies the concrete mechanism: intermodal transfer builds representational richness, the UMC progression enacts graduated process control, and PETSP-ML builds the capacity to contextualize errors, so trust calibration becomes an organic consequence of a well-designed, multi-faceted learning experience rather than an isolated add-on.

Design Implications¶

Context¶

Requirements¶

• Learning units must integrate all three ICE-T facets rather than treating them as independent strategies applied in isolation

Constraints¶

• The predicted gradient of the UMC progression is a testable hypothesis derived from the aversion literature rather than an established result; ICE-T has not been evaluated in classroom studies

Target Learners¶

• K-12 students
• students with little or no prior exposure to programming

Target Learning Goals¶

• calibrated trust in AI systems
• informed skepticism toward AI outputs
• appropriate reliance on ML systems

Affordances¶

• Ice T Didactic Framework Ml

Related Strategies¶

• Teach AI literacy as epistemic calibration and design assessment that preserves visible spaces for human reasoning and ownership
• Target calibrated trust and anxiety alongside AI literacy instruction when supporting existing generative AI users

Examples¶

-

Key Sources¶

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