---
id: "claim-sustained-ai-use-undermines-confidence"
type: "claim"
source_timestamps: ["§ Where Trust Breaks Down"]
source_url: "https://hbr.org/2026/02/how-to-foster-psychological-safety-when-ai-erodes-trust-on-your-team"
source_title: "How to Foster Psychological Safety When AI Erodes Trust on Your Team"
tags: ["confidence", "expertise", "psychological-safety"]
related: ["concept-trust-ambiguity"]
speakers: ["Jayshree Seth", "Amy C. Edmondson"]
confidence: "high"
testable: true
sources: ["adoption"]
sourceVaultSlug: "hbr-seg-adoption"
originDay: 9
articleStem: "hbr-cl-79-psychological-safety-ai-trust"
sourceUrl: "https://hbr.org/2026/02/how-to-foster-psychological-safety-when-ai-erodes-trust-on-your-team"
sourceTitle: "How to Foster Psychological Safety When AI Erodes Trust on Your Team"
---
# Sustained AI Use Undermines Professional Confidence

**Claim (confidence: high · testable).** The authors cite research showing that **sustained use of AI tools actively undermines professionals' confidence in their ability to challenge AI recommendations.**

The crucial qualification: this erosion of confidence occurs **even when the professionals possess the requisite domain expertise to know the AI is wrong.** Expertise does not inoculate against it. This phenomenon is a direct driver of [[concept-trust-ambiguity]] and shows how AI can *silently* degrade a team's psychological safety and willingness to speak up.

**How to test it:** measure the frequency and confidence of expert challenges to AI recommendations over time-on-tool, controlling for domain competence.

**Enrichment (plausible; stronger citation needed):** The extraction references "research" without naming the study. The *direction* is supported — automation-bias literature shows people fail to challenge algorithmic outputs even with relevant expertise, and Nature links AI-induced stress (mediated by psychological safety) to reduced willingness to take interpersonal risks. However, direct **longitudinal** evidence isolating *sustained use* is still limited in open sources. **Counter-nuance:** for some users AI can *strengthen* confidence (rapid feedback, freed-up time for higher-value work), so this is best read as a **risk**, not a universal outcome.


## Related across articles
- [[quote-stop-asking-why]]
- [[concept-willful-ignorance-in-ai]]