Enthusiasm in the boardroom, scepticism on the shop floor — that’s the current mood around AI. A study by staffing provider Randstad illustrates the point: worldwide, 47 percent of office employees believe AI primarily benefits their employer, not themselves — in Switzerland, the figure rises to 63 percent.
Whether AI succeeds will depend, in no small part, on whether this gap can be bridged. And that is not primarily a question of technology, but one of responsibility.
When questions of responsibility arise, the instinctive response is often to draft guidelines as quickly as possible. Put a few core values, ethical principles, and legal standards into an AI policy. Maybe launch it with a webinar. And if a company wants to signal that it takes the issue seriously, it might even set up an ethics board.
The problem is: without serious, prior engagement with the specific reality inside the company, such policies remain disconnected from everyday practice, and ethics boards struggle to fulfil their purpose. Enron, famously, had one of the most admired codes of conduct of its time. It did not prevent one of the largest corporate scandals in recent history.
This gap between paper and practice shows up with AI guidelines too: a study by a software company found that only 21 percent of employees have ever been informed about their company’s AI guidelines, and 34 percent don’t even know which AI tools are actually permitted. But guidelines that nobody knows about cannot make a difference — no matter how elegantly they’re written.
To anchor questions of responsibility, I therefore start somewhere else: my experience in sustainability consulting shows that the first step must never be a policy, but a diagnosis — through one-on-one conversations. Only where companies took the time to build understanding among employees, to clarify the organisation’s own stance — through genuine engagement, not policies pulled off the shelf — did sustainability become lived practice. Where external pressure instead forced a rushed process, the result was a pure compliance exercise; where overenthusiasm or PR calculation set the tone, it became an exercise in greenwashing.
Building Understanding
AI itself means very different things to different people — just as sustainability does.
The first step, therefore, can never be a policy. It has to be a diagnosis: building a picture of the organisation through conversations with employees that begin with broad questions and gradually become more specific. Some think of AI almost exclusively in terms of ChatGPT, while for others the horizon stretches all the way to killer robots.
From there, the questions can go further: how do employees perceive AI in their daily work, how relevant do they consider it for the company’s core, how does AI relate, in their view, to the company’s mission, and what do they understand by responsible use? Here, too, the answers can be expected to diverge sharply.
Questions like these cannot be answered — or at least not fully — through a multiple-choice questionnaire. For people to speak openly, the conversations need to be confidential and led by a neutral third party.
The point is never to convince employees of a particular view or to judge their answers. Only when people can speak openly does a realistic picture emerge.
Taken together, the interviews provide a clear picture of where uncertainty, disagreement, and hope exist within the organization. That provides the foundation for everything that follows.
It often turns out that very different realities coexist within the same company — without anyone having been aware of it until now. Making these differences visible is where the real value of the diagnosis lies. Because without a shared understanding, there is no shared stance to build on — and therefore no policy that will actually be upheld day to day, let alone a strategy that works in practice.
This lack of shared understanding also goes hand in hand with a lack of trust. The software company study mentioned earlier illustrates this well: 61 percent of executives trust AI with business-critical decisions, compared with just 9 percent of employees. Perceptions of the value of the tools in use diverge just as sharply: 88 percent of executives consider their AI solutions mature, but only 21 percent of employees agree.
And that’s no surprise, because executives and employees are often simply affected by AI in different ways. While leadership tends to benefit from efficiency gains, employees frequently bear the extra work created by checking and correcting AI output (often referred to as “workslop”). On top of that comes a more personal worry that can’t simply be argued away: fear for one’s own job. That fear can’t be dismissed with empty win-win rhetoric, nor by insisting that “there is no alternative”.
Fear of job loss has consequences: a recent study among knowledge workers found that 29 percent of respondents actively undermine their company’s AI strategy by sabotaging AI in the workplace — among Gen Z, the figure rises to 44 percent. That ranges from refusing to use the tools, to ignoring guidelines, to deliberately manipulating performance data to make AI look less effective. 76 percent of leadership already see this as a serious threat to their company’s future.
In short: with AI, hope and trust on one side meet apprehension and mistrust on the other.
From Insights to Strategy
The conclusion is clear: AI can only be integrated successfully and responsibly where a shared understanding exists — built through exactly the kind of conversations described above. Only on this basis can policies, goals, and measures be defined that don’t just sound good but actually hold up in everyday practice, because they’re grounded in the reality of the company and, above all, its employees — not in a generic template or an overly optimistic management vision.
This approach has repeatedly proven its value in my sustainability consulting work:
only a strategy that is informed by employees’ existing attitudes will stand the test of everyday practice. Tellingly, the report on knowledge workers reaches a similar conclusion: companies that involve employees in the rollout of AI and communicate openly about its intended use demonstrably reduce fear of job loss — and with it, the risk of internal resistance.
Responsibility doesn’t begin with answers but with the right questions.
Perhaps it’s time to ask those questions in your own organisation. Let’s talk. Email me, or give me a call: +41 79 292 77 55.