Four insights that will guide us into the future
Based on the ideas and discussions from the evening, we can derive four conclusions that will continue to occupy our thoughts even after the event:
Thesis 1: The most important future skill is judgment—not prompting.
Prompting can be learned relatively quickly and changes with each new generation of models. The ability to critically evaluate, contextualize, and take responsibility for AI outputs, on the other hand, remains a key future skill. Those who lack this skill not only delegate work to AI but also risk losing control over decisions.
Thesis 2: De-skilling is a risk that should be taken at least as seriously as job loss.
Figures from the World Economic Forum predict a net increase in jobs worldwide. But alongside the transformation of jobs, there is a less visible danger: that through the constant use of AI, we will lose the skills we need for our work—and may not realize this until we have to use them ourselves again.
Thesis 3: AI is only as effective at promoting learning as its design allows.
The same tool can either replace or challenge thought processes. Whether AI builds or erodes competencies in an organization is not determined by the purchase of the tool, but by the design of the surrounding processes. Simply deploying AI is not a strategy.
Thesis 4: Agility beats perfection.
In a work environment where roles, tasks, and competency requirements are constantly changing, success goes not to those who have the perfect plan, but to those who can quickly reorganize and adapt. Adaptability is becoming a critical skill for the future—for both organizations and employees.
The speakers used concrete examples and scientific findings to illustrate what lies behind these four theses and what consequences they have for education and the world of work.
The World of Work in 2030: Jobs Are Changing—and So Are the Requirements
How will we work in 2030? What tasks will AI take on? And what skills will we need to keep pace with these changes?
In her lightning talk, Dr. Sarah Genner demonstrated that technological developments not only transform existing job roles but also give rise to new occupations and forms of collaboration.
The World Economic Forum’s *Future of Jobs Report 2025* forecasts that by 2030, approximately 170 million new jobs will be created worldwide, while 92 million will be lost. The net result is an expected net gain of 78 million jobs. At the same time, approximately 22 percent of today’s jobs are affected by structural changes.
These figures are no cause for complacency. After all, newly created jobs do not automatically require the same skills as those that are being eliminated. The challenge, therefore, lies in preparing people for changing demands and giving them the opportunity to continuously develop their skills.
Technology is changing not only the nature of our work, but also the way we collaborate. New organizational structures, more flexible roles, and changing decision-making processes require companies and employees to be willing to continually question the status quo.
Why Human Skills Are Becoming More Important
The more powerful AI systems become, the more important the question becomes of what skills we ourselves need to possess.
Today, AI can conduct research, draft texts, organize information, and assist in the analysis of complex issues. However, the results must still be placed in context, critically evaluated, and accounted for.
AI can help us make decisions. The responsibility for those decisions remains with humans.
This requires in-depth expertise. After all, only those who understand the fundamentals of a topic can assess whether an AI-generated response is plausible, complete, and appropriate for the given context.
At the same time, cross-disciplinary skills are becoming increasingly important. Sarah Genner highlighted skills such as analytical thinking, creativity, resilience, a willingness to learn, and social intelligence. These are skills that help us cope with uncertainty, adopt new perspectives, and develop solutions together.
The central question, therefore, is not only how we learn to work with AI, but also how we strengthen the human abilities that make the meaningful use of AI possible in the first place.
When AI Takes Over Thinking: The Underestimated Risks
AI promises to complete tasks faster and more efficiently. But what happens when we delegate not only our work but, increasingly, our thinking to technology?
In his presentation on AI in K-12 education, Paolo R. Dettwiler highlighted four potential impacts on learning: skill-skipping, de-skilling, upskilling, and newskilling.
Two of them illustrate the risks of using AI without careful consideration.
Skill-Skipping: When Learning Steps Are Skipped
Anyone who has an AI solve a problem for them may end up with a convincing result without having gone through the necessary thinking and problem-solving steps themselves.
The learning process is cut short or does not take place at all. Skills that should develop through independent practice and understanding are not adequately developed.
De-skilling: When Existing Skills Are Lost
Those who consistently delegate analysis, formulation, or problem-solving to AI may find themselves practicing certain skills less and less frequently. In the long run, this can lead to a decline in existing skills.
Prof. Dr. Joël Krapf addressed this issue in his presentation and described a potential vicious cycle: We delegate tasks to AI, rely increasingly on its results, and contribute less and less of our own intellectual effort. As a result, skills can be gradually lost, while our dependence on AI continues to grow.
This becomes particularly problematic if, at some point, we are no longer able to assess the quality of AI results on our own.
A memorable term from Dettwiler’s lecture sums up the challenge: “incompetence-compensation competence.” This refers to the danger that people will learn to compensate for a lack of expertise using AI without first having built the necessary technical foundation.
This does not mean that AI necessarily leads to a loss of skills. Rather, it shows why we must make conscious decisions about which tasks to delegate and which skills we want to continue developing and honing on our own.
AI as a Learning Partner: How Risks Become Opportunities
The same technology that can shorten learning processes also opens up new opportunities for the targeted development of skills.
Dettwiler identified two positive developments in this regard:
Upskilling means deepening and expanding existing skills. AI can, for example, assist with research, provide personalized feedback, or offer new perspectives. However, the technical analysis and evaluation remain the responsibility of humans.
Newskilling refers to the development of new skills in conjunction with AI. This includes, for example, formulating specific queries, critically evaluating AI results, iterative problem-solving, and consciously reflecting on one’s own learning process.
So what matters is not just whether AI is used, but how.
Prof. Dr. Joël Krapf illustrated just how significant this difference can be by citing scientific studies on the use of AI tutors.
In an experiment involving high school students in a math class, the use of a conventional AI tutor did not automatically lead to better learning outcomes. A version adapted for instructional purposes was able to offset this disadvantage.
Another study involving students in a physics course showed that an AI tutor specifically designed according to learning principles can even lead to greater learning gains in less time.
The studies examined different learning settings. However, they highlight an important insight: simply providing AI is not enough. Only by consciously designing its use can we determine whether it actually supports learning.
Learning with AI: Students' Own Contributions Must Remain Visible
What does this mean, specifically, for schools, vocational training, continuing education, and everyday work?
Paolo Dettwiler made it clear that, in the age of AI, a task that promotes learning does not necessarily have to be AI-free. What is far more important is that it remains challenging and competency-based and makes students’ own thinking processes visible.
Here's an example: Students can use AI to gather ideas or get feedback on a text. However, they should still be able to explain why they chose certain arguments, accepted certain results, or rejected certain suggestions.
This requires assignments that encourage reflection, as well as clear rules for the use of AI.
The role of teachers and instructors is also changing. They are designing learning environments in which AI is used intentionally, without losing sight of the actual engagement with the subject matter.
The principle can be summarized simply:
Not every task has to be solved without AI. But students must be able to demonstrate what they have understood and accomplished on their own.
This idea can be directly applied to everyday work life. There, too, what matters should not be just how quickly a result is achieved, but whether employees understand the underlying context, can justify their decisions, and are able to take responsibility.
"Today, AI workshops are a result of convenience—tomorrow, they will be the result of a self-inflicted erosion of expertise."
Prof. Dr. Joël Krapf, Bern University of Applied Sciences, New Work Institute
What Companies Can Do Now
How can we prevent AI from weakening skills in the long term, rather than strengthening them?
Prof. Dr. Joël Krapf identified three areas of action that can also be applied to business practices.
1. Using AI Consciously
AI should not be viewed solely as an autopilot that takes over tasks as completely as possible.
As a sparring partner, she can help challenge ideas, examine arguments, and develop new solutions. It is crucial that employees critically reflect on results and actively contribute their own expertise.
2. Designing AI Tools to Support Learning
The design of the tools and work processes also plays a role.
AI applications can, for example, encourage users to verify results, document their own thoughts, or ensure transparency regarding the source of content.
The goal is not to avoid automation altogether. Rather, it is to consciously call upon human thinking in those areas where it remains essential for quality, professional development, and responsible decision-making.
3. Foster a Culture of Accountability
Companies need a culture in which AI results can be critically examined and errors can be openly addressed.
Krapf used the term “AI Workslop” to describe this: AI-generated content that, while produced quickly, creates additional work for others because it is insufficiently reviewed or poorly thought out.
This makes clear lines of responsibility, a shared understanding of quality, and a willingness to try out new ways of working and learn from them all the more important.
After all, using AI in a way that promotes learning is not just a matter of individual skills. It is also a leadership and organizational challenge.
Change is not a project with an end date
The discussion about future skills and AI raises a fundamental question: How do we support people as they navigate a constantly changing world of work?
One possible answer lies in an approach that views learning not as a completed phase, but as an ongoing process.
Three questions can help provide guidance here:
- What can I already do?
- What am I still missing?
- How do I learn that?
What sounds simple actually requires self-reflection, curiosity, a willingness to learn, and the courage to question what we take for granted.
Learn today. Shape tomorrow.
The event at Kraftwerk Zurich showed that the world of work in 2030 will not be shaped solely by technological possibilities. What will be crucial is how we empower people to use these possibilities competently, critically, and responsibly.
AI can take work off our hands, open up new perspectives, and support learning processes. However, it can also lead us to skip important steps in our thinking.
The challenge, therefore, is not to use as much or as little AI as possible. Rather, it is to structure its use in a way that allows us to maintain and further develop our skills and build new competencies.
Many thanks to the host from SKV Publishing, with its new learning universe LERN:GALAXIE, to all the speakers, the moderators, and the guests for an inspiring evening. We look forward to continuing this dialogue in a wide variety of formats and settings. Because one thing is clear: future skills, AI, and the workplace of tomorrow will challenge us and keep us on our toes.
Photography: Michelle Alexis Reber
Source:
- Presentation materials by the speakers Dr. Sarah Genner, Paolo R. Dettwiler, and Prof. Dr. Joël Krapf
- World Economic Forum, The Future of Jobs Report 2025
Frequently Asked Questions
Will AI ultimately eliminate jobs in the workplace?
According to the WEF’s *Future of Jobs Report 2025*, by 2030 there will be a net increase of approximately 78 million jobs worldwide compared to the number of jobs lost. The problem, therefore, is less the number of jobs than the nature of the change: these will be different types of work with different skill requirements. The key question is whether we can keep pace with these changes as we retrain.
Is the constant use of AI making us dumber?
It’s not inevitable, but the risk is real. Those who permanently delegate thinking and practice tasks to AI (skill-skipping, de-skilling) either fail to develop skills in the first place or lose them. If, on the other hand, AI is used as a sparring partner that challenges—rather than replaces—one’s own thinking, it can actually promote learning. The difference lies in how it’s used, not in the tool itself.
Which future skill is the most important?
Judgment. Prompting is easy to learn but becomes obsolete with every model update. The ability to contextualize, critically evaluate, and take responsibility for AI results, on the other hand, remains in constant demand and requires specialized expertise.
How can we introduce AI into our team without eroding our skills?
Through deliberate design on three levels: how we use AI (as a second opinion, not as autopilot), how we build AI tools (requiring reflection, declaring output), and how we integrate AI into the organization (addressing “AI Workshops,” assigning clear ownership). Simply providing AI is not enough. What matters is how it is integrated into processes.
