AI Readiness · Conditions
AI Does Not Create
Organizational Problems.
It Inherits Them.
The technology works. What fails is the organizational environment it deploys into - and AI has no mechanism to fix that environment. It only amplifies what is already there.
AI initiatives are widely said to fail at the same rate as major organizational change efforts, and this page used to repeat the number. It does not any more. The figure everyone quotes was traced through five separate published instances and no valid empirical evidence was found for it, which is cited below; and a failure rate means nothing without a definition of failure, which none of the sources supplies. The argument does not need the number. The explanation offered for the failures that do occur is organizational rather than technical. The technology performed as specified. The organization was not prepared to use what the technology produced.
The Central Observation
“AI does not create organizational problems. It inherits them.”
- Dan Flynn, The Four A's of Organizational Readiness™
What AI Inherits
Every organizational condition present before an AI deployment continues after it. This is not a failure of AI - it is the nature of technology in organizational systems. Technology multiplies what is already present. It does not transform the organizational environment it operates in.
Low decision velocity
An organization with slow decision-making uses AI to generate insights faster - but the insights still move through the same approval chains that produced the slow decisions. AI produces the recommendation in seconds. The recommendation waits three weeks for the governance review that was always the real bottleneck. Decision architecture determines how fast an organization can act on what it knows. AI improves what the organization knows; it does not change how fast it can act.
Leadership misalignment
An organization with misaligned leadership uses AI to execute multiple competing visions simultaneously - each individually coherent, collectively contradictory. If five senior leaders have five different definitions of what the AI program is optimizing for, the AI program executes all five definitions in parallel. The velocity of incoherence increases. What alignment problems produced before - slow progress on conflicting priorities - AI can produce faster and at greater scale.
Attention fragmentation
An organization with too many priorities adds AI as another priority. Implementation begins alongside the other eleven things currently designated as top priorities. The AI initiative receives the residual attention that remains after existing commitments are met - which is rarely sufficient for the sustained leadership focus that a complex technology deployment requires. The pilots succeed. The scaling fails. The learning from pilots never travels because no one had the protected attention to do the analysis.
Organizational friction
An organization with high coordination overhead uses AI to automate individual tasks within the existing process - without changing the process. The individual tasks execute faster. The handoffs, approvals, and coordination requirements between tasks remain. Total throughput improves modestly. The friction that limits organizational performance was structural, not task-level, and automating tasks does not address structural friction.
The Organizational Readiness Question AI Cannot Answer
The published work on AI adoption points the same way, and the sources are linked below: the predictor of value capture is not the sophistication of the AI deployed but the organizational conditions in which it operates - specifically, whether leadership is aligned on what AI is supposed to accomplish, whether decision rights are structured to act on AI-generated insights, and whether the organization has the adaptive mechanisms to learn from early AI results and adjust.
These are not AI questions. They are the Four A's of Organizational Readiness™ questions applied to an AI context. Attention: is there protected leadership capacity for this initiative, or is it competing for attention against a full existing workload? Alignment: do five senior leaders share a specific definition of what AI should accomplish and how success will be measured? Authority: does the implementation team have the decision rights to make the structural changes deployment requires? Adaptability: is there a mechanism to surface early signals of underperformance at a speed that allows correction before the investment compounds?
The diagnostic question that most AI readiness conversations skip is the one that determines whether the investment will produce returns: does this organization have the structural conditions to act on what AI tells it? If the answer is no, the path forward is not more AI investment. It is fixing the organizational conditions that will limit what AI can produce - and then deploying.
The published sources for the claims on this page, each linked, are set out below.
Frequently Asked Questions
Why do AI transformations fail?
Because of organizational conditions, not technology. Misaligned leadership, slow decision velocity, fragmented attention, and high coordination overhead all persist through AI deployment - and in many cases are amplified by it.
What organizational conditions determine AI success?
The Four A's: Attention (protected leadership time), Alignment (shared specific definition of AI success), Authority (decision rights for the implementation team), and Adaptability (mechanisms to surface and act on early signals).
What does "AI inherits organizational problems" mean in practice?
Low decision velocity stays slow. Misaligned leadership executes competing visions faster. High friction processes automate individual tasks but keep the structural friction. AI multiplies what is present - it does not transform the environment it operates in.
What should organizations do before deploying AI?
Assess the structural conditions that will determine AI outcomes: leadership alignment on what AI should accomplish, decision authority for the implementation team, protected leadership capacity, and adaptive mechanisms for early course correction.
Is Your Organization Ready for AI?
Assess the conditions before the deployment
The AI Readiness Assessment evaluates the organizational conditions that determine whether AI investment produces returns - or inherits the problems that were there before it arrived.
Not ready to run a diagnostic
Why is our AI initiative not producing value?
11 articles, in the order that answers it. No sign up, nothing to complete, and you can stop whenever the answer arrives.
- 01Organizational Readiness for AI
- 02Why AI Transformation Fails (And It's Not the Technology)
- 03The AI ROI Gap
and 8 more, ending with what to do once the constraint is named.
Follow the Preparing for AI path →What the published work actually says
Stated as each source states it, and linked, so the claim can be checked before the framework built on it is accepted.
Companies struggling to scale AI are held back by organizational and cultural barriers rather than by the technology, and the article states the point in its own subtitle: technology is not the biggest challenge, culture is.
This is the closest thing in the practitioner literature to the argument these pages make, and it is written by people who were deploying the technology rather than studying it.
Fountaine, T., McCarthy, B., & Saleh, T. (2019). Building the AI-powered organization. Harvard Business Review, 97(4), 62-73. hbr.org
Across a global executive study, the organizations getting value from AI are the ones that changed how they work so the organization and the system learn together. Adopting the technology on its own did not produce the benefit.
Adaptability, in the sense these pages use it, is the condition this report found separating the organizations that captured value from the ones that only deployed.
Ransbotham, S., Khodabandeh, S., Fehling, R., LaFountain, B., & Kiron, D. (2019). Winning with AI. MIT Sloan Management Review and Boston Consulting Group. sloanreview.mit.edu
The State of AI survey tracks adoption and enterprise-level impact separately, and the two have not moved together: most organizations report using AI somewhere, while far fewer report impact at the enterprise level.
The gap between using AI and getting anything from it is the subject of this entire site. Read the current survey rather than a number quoted from it: it is revised annually, and a figure repeated from an old edition is stale in a way the reader cannot see.
McKinsey & Company (2025). The state of AI. QuantumBlack, AI by McKinsey. mckinsey.com
The widely repeated claim that around 70 percent of organizational change initiatives fail was traced through five separate published instances of the figure. No valid and reliable empirical evidence was found to support the narrative.
So this site does not use it, and that is a deliberate refusal rather than an omission. A number nobody can source is worth less than the argument it decorates, and repeating it in front of a reader who knows its history costs more than it buys. The same test retired the AI failure statistics these pages used to carry.
Hughes, M. (2011). Do 70 per cent of all organizational change initiatives really fail?. Journal of Change Management, 11(4), 451-464. doi:10.1080/14697017.2011.630506