---
id: "contrarian-flattening-is-dangerous"
type: "contrarian-insight"
source_timestamps: ["§ The Capability-Reality Gap"]
tags: ["organizational-design", "contrarian"]
related: ["claim-flattening-orgs-risk", "entity-gartner"]
speakers: ["Julia Shin", "Sandra J. Sucher"]
challenges: "The conventional view that AI's efficiency gains should be used to eliminate middle management layers and flatten organizational structures."
sources: ["reskilling"]
sourceVaultSlug: "hbr-seg-reskilling"
originDay: 10
articleStem: "hbr-sig-50-adoption-overloading-managers"
sourceUrl: "https://hbr.org/2026/06/ai-adoption-is-overloading-your-middle-managers"
sourceTitle: "AI Adoption Is Overloading Your Middle Managers"
---
# Flattening the organization with AI is a structural risk, not a benefit

**Contrarian insight.** Conventional wisdom and major analyst predictions — notably [[entity-gartner-d50|Gartner]]'s — hold that AI's primary organizational benefit is cost savings from flattening hierarchies and eliminating middle management. The authors argue the **exact opposite**: middle managers are the critical *translation layer* for AI value, the point where junior efficiency gains and senior strategic ambitions become actual client value. Thinning this layer therefore *guarantees* AI failure; leaders must instead **over-invest** in reinforcing it. This is the argumentative core behind [[claim-flattening-orgs-risk]].

**What it challenges.** The conventional view that AI's efficiency gains should be used to eliminate middle-management layers and flatten organizational structures.

**Enrichment / counter-counterpoint.** McKinsey and Built In support the risk framing (excellent middle management becomes *more* important; eliminating layers undermines mentoring and communication). The honest other side: Gartner and some practitioners argue flattening delivers real cost and speed gains where remaining coordination work is redistributed well, and voices like 'middle management is becoming obsolete' see managers as overhead AI can absorb. Empirical outcomes likely vary by industry, firm maturity, and how coordination work is redistributed — the article describes the *risk* of naive flattening, not a universal law.