---
id: "claim-uncritical-ai-use-harms-novices"
type: "claim"
source_timestamps: ["§ How to Redesign Entry-Level Jobs", "¶14"]
tags: ["performance-metrics", "cognitive-bias", "skill-degradation"]
related: ["entity-science-journal", "concept-red-teaming-ai", "concept-microwaving-ideas"]
confidence: "high"
testable: true
speakers: ["Amy C. Edmondson", "Tomas Chamorro-Premuzic"]
sources: ["reskilling"]
sourceVaultSlug: "hbr-seg-reskilling"
originDay: 10
articleStem: "hbr-edu-46-perils-replace-entry-level"
sourceUrl: "https://hbr.org/2025/09/the-perils-of-using-ai-to-replace-entry-level-jobs"
sourceTitle: "The Perils of Using AI to Replace Entry-Level Jobs"
---
# Uncritical AI Use Degrades Novice Performance

**Claim:** A study published in the journal [[entity-science-journal|Science]] found that while generative AI can boost output by as much as **40% in text-based tasks**, novices who accept the machine's suggestions uncritically actually perform *worse* than those who reason through the problems themselves. Productivity gains from AI are therefore meaningless — and potentially harmful — if they come at the expense of developing professional judgment. This is the empirical justification for [[concept-red-teaming-ai]] and [[action-implement-red-teaming]], and the workplace parallel to [[concept-microwaving-ideas]].

**Confidence: high on the pattern; medium on the exact composite figure.** **Enrichment verification:** the general pattern is well supported — generative AI boosts speed and quantity but can harm judgment when users accept outputs uncritically, especially novices prone to automation bias. However, the specific combination of '40% output gain' *plus* 'novices who accept uncritically perform worse' appears to **synthesize findings from multiple studies** (productivity-effects work and automation-bias experiments) rather than a single, specific *Science* article. Treat the precise quantitative pairing as a composite inference; the core conceptual message — productivity gains can carry learning and judgment costs — is firmly aligned with current evidence.