---
id: "action-restrict-unstructured-inputs"
type: "action-item"
source_title: "Don't Let AI Slop Muck Up Your Company's Processes"
source_url: "https://hbr.org/2026/06/dont-let-ai-slop-muck-up-your-companys-processes"
source_timestamps: ["§ 2. Restrict the use of generative AI."]
tags: ["process-design", "hr"]
related: ["claim-policing-ai-impossible"]
speakers: ["Matthias Holweg", "Thomas H. Davenport"]
action: "Replace free-form document requests with specific, structured questionnaires to prevent AI optimization."
outcome: "Eliminates a major source of knowledge decay by forcing the submission of verifiable facts rather than AI-generated prose."
sources: ["execution"]
sourceVaultSlug: "hbr-seg-execution"
originDay: 8
articleStem: "hbr-sig-54-ai-slop-processes"
sourceUrl: "https://hbr.org/2026/06/dont-let-ai-slop-muck-up-your-companys-processes"
sourceTitle: "Don’t Let AI Slop Muck Up Your Company’s Processes"
---
# Restrict Unstructured Inputs in Processes

**Action.** Because you cannot police AI usage ([[claim-policing-ai-impossible]]), change the *format* of the information you request. Instead of allowing free-form documents (CVs, cover letters) that invite 'AI optimization,' require users to complete highly specific, structured questionnaires that capture factual information.

**Outcome.** Eliminates a major source of knowledge decay by forcing submission of verifiable facts rather than AI-generated prose, and defuses the AI-optimization arms race described in [[claim-sequential-ai-degrades-processes]].

This is **Step 2** of [[framework-four-steps-knowledge-decay]], grounded in the structured/unstructured distinction ([[prereq-structured-vs-unstructured-data]]). The enrichment overlay endorses structural constraints while noting NIST-style acceptable-use policies and content labeling as complementary partial controls.