AI adoption fails not because the technology is inadequate, but because organisations treat it as a knowledge problem when it is a behaviour problem. The SUE | Influence Framework© - developed by SUE Behavioural Design and described in my book The Art of Designing Behaviour (2024) - reveals that four psychological forces determine whether people actually change how they work. In most organisations, the blocking forces - the comfort of existing habits and the anxiety of visible failure - consistently outweigh the driving forces. Awareness training cannot fix this. Environment design can.
The Numbers That Define the Problem
A large financial services company invested heavily in Copilot licences and rolled out AI training for every department. Town halls, tutorials, champion programmes - the works. Six months later, usage data told a different story: most employees were using AI for spell-checking and summarising meeting notes. The transformative potential remained untouched. The CFO started asking questions. The project lead was quietly moved to another role.
This is not a story about a bad AI tool or an uncommitted leadership team. It is a story about behaviour. And it plays out in organisations everywhere - from Amsterdam to Singapore - whenever AI adoption is treated as a technology or communication challenge rather than a behavioural design challenge.
In my book The Art of Designing Behaviour (2024), I describe what I call de missende laag - the missing layer. Every major organisational initiative that fails to land is missing the same thing: a serious account of what actually drives and blocks human behaviour. Not what people think about the initiative. What they do on a Tuesday afternoon when no one is watching.
What the Conventional Explanation Gets Wrong
Ask most digital leaders why AI adoption stalls, and you get the same answers: employees don’t understand the benefits, leadership hasn’t communicated enough, there is resistance to change. These are not wrong observations. They are looking at the wrong level.
The problem with explaining AI adoption failure through communication or awareness is that it prescribes more of what already failed. If awareness were sufficient, the 88% of employees already using AI tools would be generating financial results. They are not - because knowing about something and integrating it into daily behaviour are two fundamentally different cognitive processes.
Behavioural science is unambiguous on this. Kahneman’s research on System 1 and System 2 thinking shows that most human behaviour operates on autopilot - fast, automatic, and anchored in existing habits. Introducing a new tool requires the brain to switch into deliberate, effortful mode. Every single time. Until the new behaviour becomes a new habit. That transition does not happen through newsletters and training slides. It happens through environment design.
The problem is not that people don’t want to change. The problem is that the environment makes the old behaviour easier than the new one.
The SUE | Influence Framework©: Diagnosing the Real Blockers
At SUE, we use the Influence Framework© - which I developed and described in The Art of Designing Behaviour (2024) - as the primary diagnostic tool before any intervention. The framework maps four forces that determine whether behaviour changes. For AI adoption, the findings are consistent across every organisation we have worked with.
Why people don’t change how they work with AI
The Influence Framework maps four forces that determine whether behaviour changes. For AI adoption, the diagnosis is clear: driving forces are real but abstract; blocking forces are immediate and emotional. This asymmetry predicts failure - and points to the fix.
The ROI gap: 88% of organisations use AI, 6% see financial results. The CFO is asking questions. The boardroom pressure is real and growing each quarter.
The “sophisticated search engine” problem: Teams are using AI for spell-checks and summaries. The transformative capability remains untouched - and leadership knows it.
Competitive anxiety: Gartner predicts 30% of AI projects will be abandoned. Nobody wants to be in that 30%. The fear of falling behind is a real force - but it is future-oriented and abstract.
Demonstrable competitive advantage: The upside of genuine AI integration is visible - but abstract and future-oriented. Difficult to feel on a Wednesday morning.
Boardroom credibility: Being the leader who made AI land would be transformative for personal reputation and organisational standing.
Regaining control: The feeling of grip - doing the right things and seeing them work - is a powerful motivator once experienced. It is rarely the reason someone opens an AI tool for the first time.
“We’re already doing enough”: The organisation has invested in licences, run training, sent communications. This creates a psychological sense of completion - even when results are absent. Sunk cost as comfort.
Familiar solution paths: More training, better tools, clearer communication - these feel right because they are what has always been done. They are also exactly what is not working.
“The problem is with the people, not the system”: When adoption stalls, attributing failure to employee resistance is psychologically easier than examining the environment that produces that resistance.
“Is this just hype?”: The fear of investing further in something that turns out to be a technology trend rather than a structural shift. This is a real anxiety, not an irrational one.
Face-loss anxiety: Acknowledging that the current approach is not working means acknowledging a mistake in front of the board, the team, the organisation. This is powerful.
“Is behavioural design not manipulation?”: A real anxiety among leaders who hear “behavioural design” and wonder whether it means tricking employees rather than helping them.
The key insight: The blocking forces are immediate, emotional, and operating at System 1 speed. The driving forces - ROI, competitive advantage, board credibility - are real but abstract and future-oriented. In every organisation we diagnose, Comforts and Anxieties win in the short term unless the environment is actively redesigned to tip the balance. This is not pessimism. It is the predictable output of a well-understood behavioural mechanism.
Three Ways This Plays Out
Scenario 1: The Training That Changed Nothing
A global technology company runs a three-day AI bootcamp for its 400 knowledge workers. Sessions are well-designed, facilitators are excellent, end-of-training feedback is high. Ninety days later, usage analytics show the needle on meaningful AI integration has not moved. Email is still written manually. Research is still done by hand. The AI tools sit in browser bookmarks, unopened.
The training gave people capability (CAN) but did nothing about the moment. No one redesigned the workflow so that opening AI was the natural first step. No one built in a SPARK - a trigger at the exact moment in the working day where AI would create immediate visible value. The capability was real. The moment was missing.
Scenario 2: The Champion Who Evangelised Alone
A large Dutch financial services organisation appoints ten AI champions - enthusiastic early adopters - and asks them to spread the word. They run demo sessions, share tips in the company newsletter, answer questions in Teams channels. Adoption in their immediate teams rises. Adoption everywhere else does not. After eight months, the champion programme is quietly discontinued.
Champions operated on WANT - personal motivation and evangelism. But WANT without CAN and AGAIN does not scale. Colleagues in other teams had no access to champions’ knowledge in the moment they needed it, and no system reinforcing the new behaviour once they had tried it once. Individual enthusiasm cannot substitute for system design.
Scenario 3: Leadership Announced, Middle Management Absorbed
A government agency launches its AI programme with a CEO statement, an all-hands presentation, and a dedicated intranet page. For the first two weeks, there is genuine interest. Then the day-to-day reasserts itself. Project deadlines, reporting cycles, client demands - the old rhythm continues, and AI integration never finds its place in it.
Middle managers - the people who structure how work actually gets done - never received a clear answer to “how does this fit into what my team does this week?” Without that answer, AI remained a leadership priority that never became a team habit. The announcement was real. The integration into the work system was absent.
Five Behavioural Interventions That Actually Work
None of the following are about raising awareness or improving communication. They are all about changing the environment so that the desired behaviour becomes the path of least resistance - which is the core principle of the SUE | Behavioural Design Method©.
-
Map the Moments That Matter First
Before any intervention, identify the three to five specific moments in each role’s working day where AI creates immediate visible value. The research meeting. The first draft of the weekly report. The preparation for a client presentation. Behavioural design begins with specificity, not with general AI capability messaging.
-
Install AI at the Point of Work (SPARK)
Remove the friction of access. AI tools that require switching applications, logging into separate platforms, or actively remembering to use them will not be used habitually. Integrate them directly into existing workflow tools - email clients, project management systems, document editors. Make the old way marginally harder than the new way.
-
Build Repetition Into Team Rituals (AGAIN)
Habit formation requires repetition in a consistent context. Build AI use into recurring team rituals - the Friday retrospective, the Monday planning session, the monthly review. Not as an add-on, but as a structural part of how work gets done. Two minutes of visible AI output per meeting is sufficient to begin the habit loop.
-
Remove the Social Risk of Getting It Wrong (CAN)
The anxiety about looking incompetent - trying a new tool in front of colleagues and getting it wrong - is a more powerful blocker than most leaders realise. Design low-stakes experimentation: dedicated sessions where failure is the point, errors are public and normalised, and no output is consequential. This removes the social barrier that keeps most people from starting.
-
Reframe: System Problem, Not People Problem
When adoption stalls, the instinctive response is to increase pressure on individuals. This triggers the anxiety forces and makes adoption worse. The reframe that works: “Our system is not yet designed for AI integration. Let’s redesign the system.” This shifts from blame to environment design - and it is the only reframe that leads to structural change.
Frequently Asked Questions
Why does AI adoption fail in most organisations?
AI adoption fails primarily because organisations treat it as a technology problem when it is a behaviour problem. They invest in tools, licences, and training - but ignore the psychological forces that keep people anchored to existing habits. The SUE Influence Framework identifies four forces: the Pains and Gains that drive people towards AI, and the Comforts and Anxieties that pull them back. When blocking forces dominate - which they do in most organisations - adoption stalls regardless of the quality of the technology.
What percentage of AI adoption projects fail?
According to McKinsey (2025), 88% of organisations are actively using AI tools but only 6% see measurable financial results. Gartner predicts that 30% of generative AI projects will be abandoned after proof of concept. These numbers suggest the failure is systemic - consistent with what behavioural science predicts when new behaviour is introduced without environment design.
What is the SUE Influence Framework and how does it apply to AI adoption?
The SUE | Influence Framework© is a diagnostic tool which I developed at SUE Behavioural Design and described in my book The Art of Designing Behaviour (2024). It maps the four forces that determine whether people change their behaviour: Pains and Gains (driving forces) versus Comforts and Anxieties (restricting forces). For AI adoption, the key finding is that restricting forces - particularly the comfort of existing workflows and the anxiety about visible failure - consistently outweigh the abstract driving forces of competitive advantage and ROI.
How is behavioural design different from change management?
Traditional change management focuses on communication, training, and achieving buy-in - all operating at the level of rational persuasion. Behavioural design operates at the level of the environment: it redesigns the situations, defaults, moments, and social contexts that trigger behaviour, so that the desired behaviour becomes easier than the old one. Rather than asking people to decide differently, behavioural design makes the right choice the path of least resistance.
What is the SWAC Tool© and how does it help with AI adoption?
The SWAC Tool© is a SUE Behavioural Design instrument for designing behaviour change interventions at Moments that Matter. SWAC stands for: Spark (the environmental trigger that initiates behaviour at the right moment), Want (making the behaviour intrinsically motivating or socially reinforced), Again (building repetition so behaviour becomes habit), and Can (removing barriers so the behaviour is actually possible). For AI adoption, this means identifying specific high-leverage moments in the working day and designing the environment so AI is the natural, available choice at those moments.
PS
At SUE, we work with organisations that have done everything right on paper: the licences, the training, the communications. And they are still stuck. The frustration is real. But so is the solution - once you stop looking at the technology and start looking at the environment in which people use it. The missing layer is always behavioural. It always has been. If you want to understand how the SUE | Behavioural Design Method© applies to your AI adoption challenge, our Fundamentals course is where we start - rated 9.7 by 5,000+ alumni from 45+ countries.
