Unless And Until We Have A Framework For Thinking About AI We Will Never Be Able To Manage It
Inspired by The Economist (September 6, 2025) and Nolan’s Six Stages of Growth
In a recent Economist article titled Normal People, the authors ask a deceptively simple question: What if artificial intelligence is just another technology? What if the current wave of anxiety and exuberance simply marks another turn in the long cycle of technological adoption?
That question invites a useful comparison with Nolan’s Six Stages of Growth Model — a framework originally developed to describe how organizations mature in their use of IT. It reminds us that every technology, no matter how disruptive it seems, passes through recognizable phases of excitement, consolidation, and normalization.
Nolan’s Six Stages of AI Growth
- Stage 1 – Initiation: The technology emerges amid wonder and fear. AI evokes both utopian visions of abundance and dystopian fears of extinction.
- Stage 2 – Contagion: Hype and experimentation dominate. Generative AI explodes into daily life, but adoption is uneven, chaotic, and often driven by curiosity rather than strategy.
- Stage 3 – Control: Organizations and regulators begin to impose discipline. Ethical frameworks, model governance, and risk management start to replace blind enthusiasm.
- Stage 4 – Integration: AI becomes part of business processes. The focus shifts from what AI can do to where it makes sense — productivity, efficiency, and augmentation rather than replacement.
- Stage 5 – Data Administration: Attention moves to quality, accountability, and lifecycle management. Questions of bias, provenance, and explainability become central.
- Stage 6 – Maturity: AI becomes normal. It is no longer treated as an existential threat or a silver bullet but as an essential, well-governed component of enterprise strategy.
From Hysteria to Normalcy
The Economist authors argue that treating AI as a normal technology leads to more grounded and productive policies — much as earlier societies learned to manage steam, electricity, or computing. That shift from exceptionalism to integration marks the difference between panic and progress.
In Nolan’s terms, we are now moving from Stage 2 (Contagion) into Stage 3 (Control) — the phase where realism, governance, and professional practice start to replace mythology.
Where Are We Now?
Most observers would agree that AI in late 2025 sits primarily in Stage 2 – Contagion. The speed of experimentation, the unevenness of results, and the public fascination with breakthroughs all fit the pattern. Yet signs of the next stages are emerging:
- Regulation and Governance (Stage 3): Governments are drafting AI safety rules, corporate boards are approving ethics frameworks, and auditors are building AI assurance practices.
- Operational Integration (Stage 4): AI copilots, forecasting agents, and customer-service bots are moving from pilot projects to production systems.
- Data Accountability (Stage 5): Growing focus on model provenance, data lineage, and documentation for AI outputs.
- Maturity (Stage 6): The first signs of normalization — AI quietly embedded in workflows, unremarkable but indispensable.
This overlap is expected. Nolan’s model does not describe isolated steps but overlapping transitions. Even while most organizations remain in Stage 2, early indicators of Stages 3–6 can coexist, forming the scaffolding for long-term institutional learning.
Predicted vs. Unexpected Behaviors
If we are indeed in Stage 2, the model would predict rapid, loosely coordinated experimentation — and that has occurred. Yet Nolan’s framework also assumed that cost constraints would quickly discipline experimentation. Instead, the cloud has made experimentation almost free, allowing Stage 2 to persist longer than in earlier technology cycles.
Dual-Cost Paradox: While the cloud has made AI experimentation nearly free for users, the underlying infrastructure is capital-intensive. As The Economist recently noted, training frontier models now requires multibillion-dollar bets by a few firms. This creates a split economy — cheap innovation at the edge, expensive consolidation at the core — allowing Stage 2 contagion to persist even as Stage 4 industrialization accelerates.
Two-Speed Maturity: Modern technologies often mature along dual tracks. In AI, the infrastructure layer — data centers, foundation models, and GPUs — is already in Nolan’s later stages, where governance and regulation are binding constraints. Meanwhile, the application layer remains in Stage 2, characterized by open experimentation and rapid diffusion. This split echoes earlier transitions such as electricity and cloud computing, where industrial consolidation at the core coexisted with creative chaos at the edge.
Takeaway and Predictions
AI maturity is not a countdown to catastrophe; it is a learning curve. On Nolan’s path from Contagion to Control, Integration, Data Administration, and Maturity, the model suggests several outcomes ahead:
- Compliance will become productized: audit logs, model registries, and safety dashboards will ship as standard features.
- Procurement will focus on performance floors and service-level reliability, not novelty.
- Data provenance will command a premium — traceable, licensed, and ethical datasets will define enterprise trust.
- Infrastructure will consolidate around a few major providers while applications diversify by domain.
- Return on investment will normalize: executives will measure AI’s value in cycle-time and cost-to-serve improvements, not hype metrics.
- AI will quietly default on: embedded in ERP, CRM, and productivity suites as an unremarkable feature of normal software evolution.
In short, the next stages are less about invention and more about integration — a slow normalization that may prove more transformative than the hype itself.
What Stage Are You In?
A quick self-check based on Nolan’s Six Stages of Growth — adapted for AI adoption and governance.
- Stage 1 – Initiation: You are exploring AI concepts and tools. Conversations are visionary, not operational.
- Stage 2 – Contagion: Experimentation is widespread but disconnected. Teams run pilots without shared governance or alignment.
- Stage 3 – Control: You have begun establishing guardrails, policies, and risk reviews. AI governance is emerging.
- Stage 4 – Integration: AI tools are embedded in production systems. ROI and reliability are your focus.
- Stage 5 – Data Administration: Data lineage, transparency, and compliance dominate. AI is part of your governed ecosystem.
- Stage 6 – Maturity: AI operates quietly in the background. Adoption follows normal change processes without hype.
Interpretation: If most statements from a single stage resonate, that is your current level. Seeing traits from multiple stages is healthy — it signals transition and learning in progress.
Further Reading: Nolan’s Six Stages of Growth
- Nolan (1973) – Managing the computer resource: a stage hypothesis, Communications of the ACM, 16(7).
- King and Kraemer (1984) – Evolution and organizational information systems: An assessment of Nolan’s stage model, Communications of the ACM, 27(5).
- Favaretto and Meirelles (2015) – Nolan’s stage level measurement of ICT/IS initiatives in modern organizations.
- Friedman (1994) – The stages model and the phases of the IS field, Journal of Information Technology, 9(3).
- SciTePress (2017) – Beyond Nolan’s Nine-Stage Model.
These sources trace the evolution of the stage model from its early mainframe context to today’s AI-driven enterprise environment — showing that technological maturity follows human and organizational learning curves more than exponential ones.
Related Reading: Something Wicked This Way Comes: A Wicked Problem