---
id: "framework-6-disciplines-gen-ai"
type: "framework"
source_timestamps: ["§ Behavioral Change", "§ Controlled Experimentation", "§ Measurement of Business Value", "§ Data Management", "§ Human Capital Development", "§ Systems Thinking"]
tags: ["organizational-capabilities", "ai-strategy"]
related: ["concept-behavioral-change-gen-ai", "concept-controlled-experimentation-ai", "concept-business-value-measurement", "concept-unstructured-data-management", "concept-human-capital-development-ai", "concept-systems-thinking-ai"]
speakers: ["Tom Davenport", "John J. Sviokla"]
steps: ["Behavioral Change: Adapting human workflows to review AI output and inject novelty.", "Controlled Experimentation: Using A/B testing to empirically prove AI's value in specific domains.", "Measurement of Business Value: Tracking ROI from individual productivity up to new product revenues.", "\\\"Data Management: Building infrastructure to capture and curate unstructured data (text", "images).\\\"", "Human Capital Development: Committing to workforce augmentation and training employees on AI skills.", "Systems Thinking: Redesigning fundamental business models to create an interlocking competitive moat."]
sources: ["spine"]
sourceVaultSlug: "hbr-seg-spine"
originDay: 1
articleStem: "hbr-cl-95-6-disciplines-genai"
sourceUrl: "https://hbr.org/2024/07/the-6-disciplines-companies-need-to-get-the-most-out-of-gen-ai"
sourceTitle: "The 6 Disciplines Companies Need to Get the Most Out of Gen AI"
---
# The 6 Disciplines for Gen AI Value

This is the **spine of the source**. Davenport and Sviokla argue that the economic value of generative AI does not come from deploying tools — it comes from building six interlocking *organizational capabilities* ("disciplines"). Firms that treat Gen AI as a technology purchase capture little; firms that master these six disciplines can achieve substantial, defensible returns.

The six disciplines, each with its own vault note:

1. **[[concept-behavioral-change-gen-ai]]** — Adapt human workflows so people know *when* to use AI, review its output for [[concept-gen-ai-hallucinations|bad predictions]], and inject [[concept-human-value-add|human novelty]]. These changes are job- and person-specific.
2. **[[concept-controlled-experimentation-ai]]** — Use A/B testing (treatment vs. control groups) to empirically prove where Gen AI actually lifts productivity or quality, rather than assuming universal gains.
3. **[[concept-business-value-measurement]]** — Track ROI rigorously, from quick individual-productivity metrics up to revenue and profit from new AI-enabled products and services.
4. **[[concept-unstructured-data-management]]** — Build the infrastructure to capture, store, and curate unstructured data (text, images, voice) that fuels Gen AI.
5. **[[concept-human-capital-development-ai]]** — Pledge augmentation over replacement, then invest heavily in AI-skills training (prompting, fact-checking, workflow integration).
6. **[[concept-systems-thinking-ai]]** — Redesign fundamental business models and interlocking processes around AI to create a competitive moat, not just localized efficiency.

**How the disciplines relate.** The first three are about *proving and extracting* value from AI in existing work; the last three are about *building the foundation and the moat*. Behavioral change and human capital development are the human-adoption pillars; controlled experimentation and business-value measurement are the evidence pillars; unstructured data management and systems thinking are the infrastructure and strategy pillars.

**Mastering the disciplines is necessary but not sufficient.** The authors pair this framework with a second one — [[framework-gen-ai-project-selection]] — that governs *which* Gen AI projects to fund so the capabilities translate into realized value.

Enrichment note: McKinsey, BCG, and Bain independently emphasize a very similar capability set (change management, experimentation, data, skills, operating-model redesign), indicating this framework aligns with broad expert consensus rather than being idiosyncratic. Both authors — [[entity-tom-davenport]] and [[entity-john-j-sviokla]] — are the source of this framework.


## Related across articles
- [[framework-value-creation-pyramid]]
- [[framework-5-types-ai-investment]]