By Grzegorz Leszczyński, Piotr Gaczek, and Andreas Munzel
Your team doesn’t have an AI adoption problem – it has four different psychological profiles toward AI, and only 7% are true believers.
Walk into any leadership meeting today and you will find broad agreement that artificial intelligence matters. Walk out of it, and you will find that agreement evaporates the moment individual managers return to their desks. One quietly runs every decision through an AI assistant. Another uses it to draft emails and nothing more. A third avoids it entirely, convinced it threatens both quality and jobs.
Most organisations treat this as a training problem or, worse, a compliance problem: roll out the tools, run a workshop, expect uniform adoption. Our research suggests this approach is doomed from the start — because managers do not differ merely in how much they use AI, but in how they fundamentally understand what collaborating with it means.
In a study recently published in Technovation, we surveyed 472 marketing managers in the United States about how they work with AI in their decision-making. Rather than sorting them by usage frequency or technical skill, we examined the deeper structure of their collaboration: whether they use AI to go deeper within their own domain, to bring in perspectives from other business functions, or to genuinely integrate AI-generated insights into how their teams think and decide. We combined this with their sense of control over AI, how transparent they find it, how their organisation supports experimentation, and how much decision-making authority they are willing to share with algorithms.
The analysis revealed four distinct profiles. Recognising them in your own organisation is the first step towards getting everyone onto the same page.
The four profiles
AI Trailblazers (7% of managers). A small vanguard that treats AI as a genuine strategic partner. They use it intensively within and across business functions, feel firmly in control of it, and understand how it reaches its conclusions. Strikingly, they are the only group prepared to make a substantial shift in decision authority: they would accept AI playing a role in nearly half of their decisions, up from about a third today. Their organisations actively encourage AI experimentation — and these managers report the highest satisfaction with market performance of any group.
AI Strategists (41%). The disciplined adopters. They value AI highly and use it broadly, but always within a governance frame: AI as a co-pilot in a human-led process, not an autonomous actor. One Strategist in our study put it plainly — AI speeds up decision-making dramatically, but always as part of a process where humans lead. They favour structured, incremental expansion of AI’s role rather than bold leaps.
Pragmatic Adopters (42%). The largest group. They use AI where it demonstrably helps — analysing data faster, drafting content, optimising campaigns — but keep it in a service role. They describe AI as a “second pair of eyes” whose output still requires human quality checks. Their sense of control and their understanding of how AI works are middling, which is precisely why they keep their engagement task-specific rather than strategic.
AI Skeptics (10%). The holdouts. They see little value in AI as either partner or tool, feel the least in control of it, and find it the least transparent. Their concerns are not irrational contrarianism: they voice genuine unease about job displacement and the erosion of human creativity and judgement. Notably, they also work in the least AI-supportive organisations — and report the lowest satisfaction with market performance.
Why this matters more than adoption rates
Three findings should give leaders pause.
First, the enthusiasts are rarer than you think. Trailblazers make up just one manager in fifteen. The overwhelming majority of your management team — more than 80% — sits in the cautious middle, using AI selectively and insisting on human primacy. If your AI strategy is designed around the assumption that managers are eager to hand work to algorithms, it is designed for a group that barely exists.
Second, attitudes towards AI are not really about the technology. What separates the profiles is not technical skill but psychology: the felt sense of control over AI, and the degree to which its reasoning seems understandable. Skeptics do not resist AI because they cannot operate it; they resist it because it feels opaque and beyond their influence. This means the lever for change is not more product training but more transparency and more genuine human agency in how AI is deployed.
Third, organisational climate and performance move together. Managers in organisations with a proactive stance towards AI — a climate that encourages trying new ways of using it — report significantly higher satisfaction with their market performance. We cannot claim simple causality from these data, but the pattern is consistent: where experimentation is safe, collaboration matures, and confidence in results follows.
What leaders should do
The central managerial implication is that one-size-fits-all AI programmes waste money on some managers and alienate others. Each profile needs something different.
Deploy your Trailblazers as multipliers, not outliers. Give them room to experiment — AI strategy labs, pilots of advanced tools — and formalise their role as mentors. Their enthusiasm is contagious, but only if the organisation channels it rather than leaving them as isolated power users. One caution: their expectations of the technology can outrun what it currently delivers, so pair their ambition with honest evaluation.
Give your Strategists governance, not cheerleading. This group will expand AI’s role willingly if the guardrails are explicit: clear human oversight points in decision processes, scenario-based training on when to trust AI and when to override it, and explanations tied to real business cases. Slogans about “AI transformation” will not move them; well-designed processes will.
Win over Pragmatic Adopters with proof, not vision. Show them concrete, task-level value — campaign optimisation, lead scoring, reporting automation — with simple dashboards that make AI’s reasoning legible. Let them control how much influence AI has. Each demonstrated win expands their comfort zone incrementally.
Take your Skeptics seriously. Their fears about displacement and lost control are the very concerns that responsible AI governance exists to address. Start them with low-stakes analytical tasks, guarantee human-in-the-loop defaults and manual overrides, and pair them with trusted, AI-positive peers rather than official evangelists. And resist the temptation to see them purely as a problem: their vigilance about data quality, opacity, and overreliance is a governance asset that unbridled enthusiasm lacks.
The real message
The question facing organisations is no longer whether managers will work with AI, but on what terms. Those terms are being negotiated individually, desk by desk, shaped by each manager’s sense of control, understanding, and trust. Leaders who diagnose these orientations — and meet each group where it stands — will integrate AI faster and more durably than those who simply mandate adoption. In human–AI collaboration, as in any collaboration, the relationship comes first.
Grzegorz Leszczyński
Piotr Gaczek
Andreas Munzel
