By Nadine Hughes, Engagement Delivery Director - Salesforce, Australia
The scale of AI investment is unprecedented, yet failure rates are soaring. A 2025 MIT report found that 95% of enterprise generative AI pilots fail to scale, while RAND Corporation puts the overall AI failure rate at over 80%, double that of non-AI programs. These aren't just growing pains; they represent a systemic misunderstanding of the problem. When adoption plateaus, leaders typically blame the technology: the wrong tool or implementation partner. However, the real cause goes unexamined. AI transformation fails primarily because it is misdiagnozed as a deployment challenge rather than a fundamental behavior change challenge. Behavior change is a science, yet it is rarely applied to AI adoption. I have developed The CORE Model™- Clarity, Ownership, Rewire, Embed; to close this gap. Drawing on 24 years of transformation leadership and psychology, it addresses the human system, not just the technical one.
Why AI Is Not Like Other Technology Programmes
AI failure is categorically different from standard IT projects, which fail at rates of 25-50%. Research shows 61% of failed AI projects treated the initiative as an IT project rather than a business transformation. This conflation ignores how AI uniquely impacts people. Conventional technology changes how people work while keeping professional identity intact. AI is different for four psychological reasons: it challenges expertise and identity; it is non-predictable, requiring a different kind of trust; the pace leads to change saturation; and the emotional stakes are higher, typically fear of losing jobs. The dominant change management sequence of selection, training, and measurement underestimates this complexity. Resistance is not an obstacle to manage, but a symptom of psychological misalignment.
CORE Model™: A Psychologically Grounded Framework
The CORE Model™ treats AI adoption as a behavior change problem. Its sequential stages address distinct psychological barriers using research in motivation and habit formation. It moves from internal diagnostics to externalized cultural defaults, ensuring that the technology is not just installed, but integrated into the cognitive workflow of the organization.
Clarity — Diagnosing the Fear Landscape
Clarity is diagnostic. Leaders must surface what is genuinely driving resistance, which is often masked by technical questions. Drawing on Amy Edmondson’s work on psychological safety, this stage requires creating environments where employees feel safe to express identity-based anxieties. Without mapping this fear landscape, all subsequent interventions are built on misdiagnosis. When workers feel their expertise is threatened by an algorithm, they do not need more training; they need the psychological safety to explore their new role without fear of retribution or obsolescence.
Ownership — Building Commitment Through Autonomy
Ownership moves beyond brittle compliance. Mandated rollouts often fail once leadership attention shifts because the change wasn't internalized. Based on Self-Determination Theory, durable motivation requires autonomy, competence, and relatedness. Ownership is achieved through genuine co-design, where employee input materially shapes the integration. When people see their own insights reflected in the AI’s implementation, they move from being passive recipients to active advocates.
Rewire — Environmental Design and Habit Formation
Rewire acknowledges that motivation alone is insufficient for sustained change. Behavior change is substitution, not addition. As established by BJ Fogg and James Clear, the most reliable way to change behavior is to redesign the environment so the new pattern is the path of least resistance. Willpower is not a strategy; leaders must remove friction from AI processes and make regression inconvenient. We must "stack" AI habits onto existing triggers, ensuring the new tools become the default response to routine tasks.
Embed — Cultural Integration and Longevity
Embed asks if change outlasts the program. Change that depends on CEO momentum is managed, not embedded. Analysis shows 56% of failed AI projects lose C-suite sponsorship within six months. Embedding builds new practices into the organizational culture and default systems, as defined by Edgar Schein. Success is measured by capability persisting 24 months after launch, where AI is no longer a "project" but simply "how we do things here."
Trust as the Through-Line
The CORE Model™ treats trust not as a checkbox, but as the medium through which change travels. Each stage has an operative dimension: trust in the situation (Clarity), trust in leadership (Ownership), trust in oneself (Rewire), and trust in the system (Embed). When AI fails, it is usually because one of these human dimensions collapsed. Trust is the lubricant that reduces the friction of the four unique psychological barriers AI presents.
The Business Imperative
Failed AI projects cost trillions globally and erode workforce trust. Competitive advantage belongs to organizations that build human capability alongside technical tools. The CORE Model™ provides a framework for this, applying decades of behavior science to the unique psychological challenge of AI. By addressing Clarity, Ownership, Rewire, and Embed, leaders can finally bridge the gap between pilot and production. The field needs better thinking about people, not more technology frameworks. Organizations that master the human side of AI will be the only ones left standing in the next decade.
