Founders Talk #02: Rethinking L&D with Generative AI

Question 1: Which practical implementation of AI has impressed you the most so far?

Marius : From a business perspective, I would highlight the Klarna AI case. A few months ago, Klarna announced that they can now handle almost two-thirds of all service chats via a GPT variant - and that's actually without a human in the loop. To see what is already possible on the voice and video side is really impressive. Figure's partnership with OpenAI is also exciting. Some time ago, Figure released a video showing a robot that can have conversations, argue, and manipulate objects in real time - and it just looks very "human".

David : Another fascinating use case is Devin. This is an AI software engineer that can simulate a complete developer. It can independently search for solutions and build complete chains of reasoning. In fact, I believe that this "agent case" will serve as a core technology in many recurring task packages in the future, not only in learning and development, but in other areas as well.

Question 2: Do you have a few examples of how you as a manager already use AI in your day-to-day work?

Marius : For me, there are three main areas of application. The first is speech AI, which I use the most. It helps me enter my thoughts into the system and structure them beforehand so that I can continue working with a well-organized text version. Second, I appreciate it when the AI takes on the role of "devil's advocate" and shows me where my mistakes are and what is still unclear. And third, of course, is learning. The AI analyzes in advance which important building blocks are needed to learn a skill and checks if you have really understood that skill.

David : For me, the use cases can be roughly divided into two categories. The first is brainstorming, i.e. overcoming the blank page problem. For example, if I have to prepare a presentation on a certain topic or write a product memo, it often helps me if I have already pre-formulated the obvious core statements.

The second category, which is becoming more and more exciting for me, is aggregation. This is about summarizing the collected information in a concise headline or finding the right metaphor that expresses the essence. GPT and generative AI in general have helped me enormously with this aggregation work recently.

Question 3: What skills are most companies lacking for the move towards AI?

David : This is an extremely exciting question that not only concerns Learning and Development departments, but can be relevant for every single person in the long term. Generative AI is a so-called general purpose technology that actually affects all areas.

We often see that the use of AI often starts with managers, who play a central role. They need to understand the basics of AI, apply it themselves, and develop the skills to inspire the rest of their organization. When it comes to AI, there are often meta-skills that are needed and often lacking: the ability to understand, apply, and try AI, the courage to try new things, the foresight, and the ability to inspire others. It is equally important to know the limits of AI and to ensure that managers actively use the technology. After all, they have to pass this knowledge on to their employees.

Last but not least, what is missing at employee level: above all, the courage to disrupt your own work and question: where can AI help me to become better or more efficient? To look at my own areas of responsibility and see where AI can already support me today.

Question 4: And now in concrete terms: what steps are necessary to implement the various AI use cases?

David: It always starts at the top, i.e. with the C-level managers, who have to be open to new processes and business models. This is the absolute foundation for any innovative work process.

The second step is to familiarize employees and managers in particular with the basics. This includes the theory: What is generative AI? What can it do and what can't it do? At the same time, the right tools need to be provided so that employees can use these technologies in the first place.

The last major module is upskilling in the area of ethics and governance. This is about what we as companies and managers need to consider in order to use AI for the common good and minimize risks.

Question 5: And now to L&D. What do you think AI means in the area of L&D?

Marius: For our L&D administrators, there are a number of recurring work processes in which we can already make excellent use of generative AI. This starts with classic activities such as maintaining metadata. Whenever we need to create descriptions, be it for new course descriptions or image captions, AI can take over these tasks so that we no longer have to intervene manually.

Building on this, we can move on to more complex tasks that require more context. A common example is the concept of skill tagging. We have unstructured data from training catalogs that need to be linked to skills from a skill taxonomy. I believe that skill management will be a fundamental use case that is much easier today than in the past.

When I listen to the stories of customers who tried to establish a skill taxonomy in their company five to ten years ago, the effort and budget for such activities is six to ten times less today. One of the biggest challenges was initially creating a basis of company-specific skills and identifying duplicates. This was particularly challenging in cases where we needed to understand the context, i.e. words that look different but have the same meaning. This is exactly what we did for major customers such as Deutsche Bahn.

Another important area of application is the creation of learning materials. This is often the best-known use case. However, it becomes more exciting when we know which materials we need to create. Based on the training catalog and the comparison with our skills catalog, we can see where our gaps are and which materials should be created or provided in the short term because we have identified important skills that we do not currently cover in our company.

Question 6: What do you think the content providers of the future will look like?

David : In the future, content providers could act as a kind of content generation engine, based on specific data sets and oriented towards didactic experiences or ideas. What makes these companies special is no longer the better camera, the better setting or the more impressive studios, but the systems and agents in the background that can produce the content. I believe that there will still be content providers, but they will no longer be the traditional content producers we know today.

Question 7: And what do you think about the impact of AI on learners and learning?

David : After much thought, many conversations and experimentation, it was clear to us as a product team at edyoucated that the true transformation is less about the creation of materials and more about the navigation and “how” of learning.

Initially, this means: how do I find the next best course for me and how do I learn the next best skill? We often call this macro-personalization. This means that the AI can suggest the best possible course or skill for me based on my current level of knowledge and my goals. Then, of course, there is micro-personalization, which takes place at course level, i.e. adaptive learning paths. This is not new, but it is crucial in order to provide learners with the most efficient path to their goal. And only then, for me, does the big AI revolution in learning materials come.

However, there is one specific use case at the level of learning materials that I am very convinced of: simulations. For example, a chat-based AI is used to train customer service by taking on the role of a disgruntled customer. This has several positive effects. Firstly, no direct sparring partner is required, which saves time and costs. Secondly, learners are motivated to try things out and are more open to making mistakes. Learning by doing is still the best way to learn.

Marius : What I also find particularly exciting, alongside the classic learning formats and simulations, is the idea of zero-cost tutoring. It has been proven that one-to-one support has a lasting positive effect on learning success. This is exactly where generative AI can come in and offer agile tutoring at eye level based on collected data. The system knows what I have just looked at, what I have learned so far and what my goals are.

This enables formal training based on practical examples. I believe this will have a dramatic impact on what is possible in companies. I personally believe that this will be one of the main features that Learning Management Systems will include as standard in 5 to 10 years.

Question 8: How do you think the role of an L&D manager will change in the future?

Marius: The role of L&D teams will increasingly evolve into that of a business partner who supports strategic decisions and adapts skills to the corporate strategy. Traditional L&D managers will be less in demand as the focus shifts from didactic and training-oriented tasks to data-driven and strategic positions.

The use of AI enables a more transparent organization in which skills and tasks can be better aligned, leading to more efficient workflows. Future L&D teams will focus more on data-driven decisions and perform fewer manual tasks. An example of this would be the transition from creating presentations to more intelligent solutions that provide information on demand. Ongoing data analysis will be replaced by AI-driven tools that actively inform decision makers and display relevant information, leading to better decision making.

Thank you for your time and we look forward to the next Founder Interview.