Artificial intelligence is currently the dominant topic in businesses. Expectations are huge, as is the uncertainty. What does AI mean for leadership? Which tasks will remain the preserve of humans? And why do so many AI projects fail despite enormous investment?
I discussed this (in our latest podcast, in German language) with the sociologist of work, Prof. Dr Tobias Kämpf, who has been researching the digitalisation of the world of work for almost twenty years.
Surprisingly, our conversation focused very little on technology – and almost entirely on people. That is probably the most important insight.
AI is not an IT project
When asked what managers should be doing now when it comes to AI, Tobias replied without hesitation: “The most important factor for success is the involvement and empowerment of ordinary employees.” This is because, in many companies, the introduction of AI is still a top-down process: senior management decides, IT implements, and employees only find out late in the process what will change for them. The problem is that AI doesn’t just change software; it changes work.
Why so much knowledge is lost
Drawing on his interviews – for example, with call centre staff – Tobias Kämpf reported: “How can anyone come up with the idea of building a chatbot without having spoken to the people who have been on the phone with customers every day for the past ten years?” This is precisely where the fallacy lies: anyone who speaks to customers or manages processes on a daily basis knows where processes break down, what exceptions arise and why some ‘perfect’ solutions never work in practice. This knowledge often determines whether a solution is accepted. That is why Tobias Kämpf does not view the introduction of AI as an IT project, but as a collaborative development process.
Case study: A bank wanted to automate its mortgage lending process. Land registry extracts, notarial deeds and building plans were to be processed using AI. Simple in theory, but complex in practice: every land registry office and every notary worked differently. The solution: IT and specialist departments worked together in interdisciplinary teams. Initially, things were rather bumpy, as both sides first had to learn each other’s language. Yet it was precisely this mutual learning that ultimately led to a successful outcome.
The ChatGPT moment – a guerrilla innovation
The success of generative AI often seems as though it was strategically planned. But it wasn’t. “Employees realised in their private lives: ‘Wow, these tools are so useful,’ and brought them into the company on their personal devices,” says Tobias Kämpf. Not management, not IT – the employees themselves, often long before any official policy was in place. Tobias calls this ‘grassroots or guerrilla innovation’. Innovation doesn’t arise where it’s planned, but where people have the freedom to experiment. This fundamentally changes the role of leadership: away from providing answers, towards asking questions and listening. Because nobody – not even management – can know all the answers anymore.
AI projects rarely fail because of the technology
“We think we can introduce the technology as if it were a plug-and-play tool, and that it will solve all our problems. That is the fatal mistake,” says Tobias Kämpf. New AI does not replace poor processes, nor does it create collaboration on its own. The traditional interplay between people, technology and organisation remains crucial. Anyone who merely changes the technology will fail because of the other two factors.
Change management begins much earlier
Rather than starting with training only after software development has been completed, Kämpf believes that change begins the moment people are – or are not – involved in the design process. To achieve this, managers need to provide fewer answers and create more space in which people can experiment and learn. They need a basic understanding of the technology in order to realistically assess its opportunities and limitations. More importantly, however, they must enable change rather than control it. “It’s about creating spaces where people can experiment,” says Tobias Kämpf. After all, nobody knows today which applications will become established. Experimentation and collaborative learning are becoming central leadership tasks.
The change primarily affects the managers themselves
When it comes to change management, we usually think first of all of the employees. Yet the work of managers is also undergoing fundamental change. Decisions are prepared on the basis of data, whilst routine tasks are disappearing. Tobias Kämpf spoke of experienced engineers in the automotive industry who were suddenly expected to develop software: “Professional identities have to be redefined.” This applies just as much to managers, who in future will need to focus more on facilitating, networking and providing guidance rather than simply drawing on their specialist knowledge.
AI can make work more human
Instead of asking, “Which AI should we buy?”, we should be asking: Where do employees waste time unnecessarily every day? Tobias Kämpf himself was sitting over his travel expense claim after a train journey and thought: “It’s a shame I don’t have an agent for this yet.” His recommendation: “You should ask employees: ‘Where are your pain points?’” Because the focus isn’t on the technology, but on day-to-day work.
Productivity isn’t automatically a bad thing
In many companies, productivity has negative connotations, evoking images of staff cuts and pressure. Tobias Kämpf sees it differently: what matters is who benefits from productivity gains. If they merely lead to cost reductions, they won’t generate any enthusiasm. If, however, they reduce routine work and create space for customers and more creative tasks, AI becomes a genuine asset. But: “It’s not enough to build acceptance. We need to generate enthusiasm,” says Tobias Kämpf. Acceptance means “all right then, so be it…”. Enthusiasm means “Great, that really does help us!”. People get involved when they recognise a clear benefit for their own work from that involvement.
We now have the opportunity to reshape work
Tobias Kämpf drew a historical parallel: the ‘Humanisation of Work’ programme of the 1970s sought to improve both production and working conditions – an idea that is more relevant today than ever. He concluded by saying: “We are in the midst of a profound transformation of work and the economy.” This is comparable to industrialisation, but this time it concerns knowledge work. The difference: “Because many structures are currently in flux, they can now be changed.” If companies fail to seize this opportunity, there is a risk of uncertainty and social tensions: “We must shape this positively; otherwise, social cohesion and unity in the world of work will be seriously at risk.”
Conclusion
The real challenge of AI is not the technology – it is leadership. Companies need fewer people who believe they know all the answers, and more leaders who ask good questions and create space for learning and active participation. The success of AI is rarely decided in the server room, but rather in the meeting room – specifically, when leaders ask: “What would really make your work better?”
Best regards,
You can also find our podcast on Spotify and Apple Podcasts.
You can subscribe to our newsletter here.
graphic: flow consulting
