---
id: "concept-hidden-substitution"
type: "concept"
source_timestamps: ["§ The Hidden Substitution"]
tags: ["automation-risks", "incentives"]
related: ["concept-implicit-organization", "claim-deleting-motivational-mechanisms"]
definition: "The inadvertent replacement of complex human motivational and social accountability mechanisms with narrow, formalized AI optimization functions."
speakers: ["K. Sudhir"]
sources: ["agentic"]
sourceVaultSlug: "hbr-seg-agentic"
originDay: 6
articleStem: "hbr-new-26-agentic-systems-implicit-rules"
sourceUrl: "https://hbr.org/2026/06/how-to-design-agentic-systems-around-the-implicit-rules-that-govern-your-company"
sourceTitle: "How to Design Agentic Systems Around the Implicit Rules that Govern Your Company"
---
# The Hidden Substitution

**Definition:** The inadvertent replacement of complex human motivational and social accountability mechanisms with narrow, formalized AI optimization functions.

The hidden substitution is the phenomenon where treating AI agent deployment as a mere 'process improvement' (same workflow, faster execution) quietly deletes the web of motivational mechanisms that governed human workers.

**Worked example — the underwriter:** A human underwriter balances approval speed against reputational risk and social accountability (facing colleagues after a bad loan). An AI underwriter optimizing for speed and historical default rates lacks these social and career stakes, replacing a complex, consequence-bearing human judgment matrix with a brittle, narrow optimization function.

This substitution is 'hidden' precisely because the visible workflow looks unchanged — only the invisible [[concept-implicit-organization]] has been removed. The consequence is captured in [[claim-deleting-motivational-mechanisms]].

**Enrichment note (tension):** A counter-view holds that motivation and accountability need not simply vanish — organizations can *re-architect* incentives around AI systems (tying system performance to human managers' evaluations, audit trails, embedding normative constraints in objective functions). The risk is real *if* incentive redesign is ignored, but deletion is not inevitable.