Teach AI literacy — how and when LLMs fail — to guard against cognitive surrender¶

Strategy · All strategies
Evidence · no claims cited

Description¶

Among the directions the article opens is AI literacy: "whether teaching users how and when LLMs fail can restore judgment suspension and, more generally, guard against cognitive surrender". The strategy is to instruct learners about the failure modes of large language models so they retain the metacognitive option of saying "I don't know" when consulting or receiving AI advice. The article poses this as an open empirical question rather than a tested intervention.

Design Implications¶

Context¶

Requirements¶

• Instruction that makes LLM failure modes visible to users before or while they receive AI advice

Constraints¶

• The article has not tested this strategy; it names it as a future research direction
• Whether larger or reputational stakes would close more of the suspension gap is unknown

Target Learners¶

• Adult AI users in education and workplace settings

Target Learning Goals¶

• Restoring willingness to suspend judgment under uncertainty
• Guarding against cognitive surrender to fluent AI output

Affordances¶

• Epistemia Fluency Threshold Account

Related Strategies¶

Examples¶

-

Key Sources¶

• Chiara Marcoccia, Walter Quattrociocchi, Valerio Capraro. (2026). AI advice suppresses people's willingness to say "I don't know", even when the advice is wrong and accuracy is incentivized. https://osf.io/nwx6r/overview?view_only=a03ae0ca587c4c84a442c44d55ab97b6