---
id: "evidence-pmc-collaboration-cwb"
type: "evidence"
source_timestamps: ["Enrichment: Adjacent Literature [2]"]
tags: ["external-evidence", "emotional-fatigue", "collaboration-design"]
related: ["concept-ai-brain-fry", "claim-brain-fry-errors", "claim-quality-control-decline"]
source_org: "PubMed Central (PMC)"
citation: "PMC-hosted study — employee–AI collaboration increases loneliness and emotional fatigue, which predict counterproductive work behavior."
supports: ["concept-ai-brain-fry", "claim-brain-fry-errors"]
validation_role: "supporting (directional)"
sources: ["agentic"]
sourceVaultSlug: "hbr-seg-agentic"
originDay: 6
articleStem: "hbr-ext-16-dont-treat-agents-like-employees"
sourceUrl: "https://hbr.org/2026/05/research-why-you-shouldnt-treat-ai-agents-like-employees"
sourceTitle: "Research: Why You Shouldn’t Treat AI Agents Like Employees"
---
# Employee–AI Collaboration → Loneliness, Fatigue, Counterproductive Behavior (PMC)

**External evidence (enrichment overlay).** A **PMC / PubMed Central** study found that collaboration with AI can increase **loneliness** and **emotional fatigue**, which in turn predict **counterproductive work behavior (CWB)**.

**How it relates to this vault:** It supports the idea that **collaboration design matters, not just model capability** — the same principle driving the source's rejection of naive 'AI teammate' framing. It gives adjacent grounding for [[concept-ai-brain-fry]] and the error-rate spikes in [[claim-brain-fry-errors]], and for the reduced scrutiny in [[claim-quality-control-decline]]. It also suggests candidate metrics for the open question [[question-measuring-brain-fry]] (fatigue as a leading indicator of error/CWB).