VG Wort Zählmarke
3 AI myths from the SME sector
13:50

Skepticism and reservations, fascination and anticipation: no matter how you feel about AI – you are certainly not alone. No wonder, says Mario Trapp. After all, AI is a pioneering technology. It gives rise to tangible changes. And it provokes. Even myths: “We’re in the SME sector here. AI is too far removed from what we do at the core," is one assumption. A false one, says Trapp. As a Partner in Restructuring & Transformation and Head of Digital Strategy, he has been observing the technological developments of AI for quite some time. His world? Big data and data analysis. His industries? Many – primarily owner-managed companies. How does he himself apply AI in business? Should companies already be hiring Chief AI Officers? And which AI developments make even him uneasy? Get to know our new partner. And how to improve your prompts.

 

Do you still remember your first prompt, Mr. Trapp?

Yes. “How do I make good use of eight hours in Singapore?” I had a layover there. Instead of reading a travel blog, I asked ChatGPT. Surprisingly, it provided some really good tips in the first round. So I kept on tweaking: “I’m interested in culinary delights and lively neighborhoods. I would like to visit as many sights as possible. All of this should be optimized in terms of time and route and using public transport.”

E voilà?

It worked wonderfully. Even specific timeslots for immigration were recommended to me. To be on the safe side, I validated the results: Everything was plausible. Since then, I’ve never planned a trip without AI. An exciting use case.

Indeed. But you also need to be able to prompt well. What are the key characteristics of a good prompt?

A good prompt is clear, precise and contextual. It contains a specific request for action without any digressions or nested sentences. The task is clearly formulated, for example: “I have problem X in area Y. It has existed since period Z and is characterized by the following symptoms. Please explain possible solutions for problem X, taking factor Y into account.” This guides the recipient to respond in a targeted and effective way. You can also include a hypothesis. A good prompt contributes to better categorization and ultimately also provides more precise answers.

And is that it?

Not by a long shot. There are many different strategies. Sometimes it is helpful to word prompts in such a way that the answers generated outline comprehensible solutions. This is called “chain of thought prompting.” The aim is to understand why the AI outputs this particular solution path or result. Which is very interesting in the context of formulas and codes. In general, the field of “prompt engineering” is an entire science. And it is currently picking up speed.

If it all works so well: Do you also use AI in your professional practice?

Yes, although the challenge in business is that we have to be very careful in terms of data protection about which data we make available to which AI model. Nevertheless, there are a lot of cases where AI can be used. And I’m not talking so much about texts – we’ve all tried it and found out: The results turn out good, but at the same time they’re quite generic. What I find much more exciting – from the perspective of a consultant and data analyst – is that certain problems, such as creating complicated formulas or codes to analyze a data set, can easily be solved using AI. So instead of diving into the depths of the web, searching for content from documentation, repositories and forums – which can take a long time – I describe the problem to the AI, ask for a solution, a formula, a snippet of code and find that I get amazingly good answers. That is point 1.

Point 2?

Point 2 is exactly the other way around and just as exciting: I look at the code snippet or a formula, but perhaps I don’t understand the author’s idea, or at least not completely. My prompt here is: “Explain the structure and logic of this code or formula to me: How does the construct work in detail? Why is this a good way? Is there a better way?” And by means of “reverse engineering” – that’s the reverse process – I get an explanation from the AI that would otherwise have taken me a long time to find. Plus: in some cases the AI provides a report on performance issues. Based on this, I can then optimize the formula or the code.

Sounds fascinating. You studied business administration and business informatics with a focus on big data and data analysis. Your hobbyhorse is advanced data analysis.

That’s right. Data plays an important role in my career. Where Excel ends is where it gets interesting for me.

What do you find so exciting about it? Why do you focus on data of all things?

For me, the question of why has always been very essential. To penetrate things in depth and develop an understanding of their interrelationships. That certainly helped me to choose this career path. Ultimately, consulting is always about problem-solving and preventive work. To do this, however, first you need to clarify the facts. Any relevant information, which can often relate to quality, helps here. However, it does not always enable an in-depth analysis of the problem.

But data does?

Most of the time. This is because data is generated wherever systems record movements and states. This is now the case almost everywhere. Data makes it possible to understand at a very granular level what has actually happened and where it has led. Data can be used to recognize patterns. Plus which, and this is something that I also find very fascinating: all experts – regardless of their focus and experience, regardless of their industry – meet on the basis of pure facts at the level of data.

If consultancy is your first home, what would be your second?

Medicine, I think. I would have become a doctor.

What other specific questions guide you in your work – apart from the why?

How does value flow through a company? Where and how is value created – but where is it also destroyed? Where can I have a positive influence? And what is the best possible result in the end?

You’ve been with enomyc for ten years now, a Partner in the Restructuring & Transformation division and Head of Digital Strategy. How long have you had AI on your radar? And did you realize straight away that it would also come to play an important role in consulting?

I have been following developments for a long time, partly via trend monitors such as the Gartner Hype Cycle. It had been apparent for some time that various AI technologies were in the pipeline. And yes, I saw the potential for consulting in this respect as well. Nevertheless, for a long time I asked myself: What is the specific use case for this or that AI? And where exactly can these forms be used? In some cases, this was still a long way off. Until in the last two to three years – and I think everyone has noticed this – there was a big turnaround: AI is here to stay. There’ll be no getting around it. And it brings very tangible benefits beyond business consulting.

What is the most frequent prompt that companies ask you about AI?

I think the analogy is very good, because orders are prompts. A classic prompt is: “How far should we allow AI into the processes of our business model?” or “What decision-making authority should we place in AI?”.

Are you already advising some companies to create positions such as “Chief AI Officer”?

In some cases. I don’t think it’s absolutely necessary to install AI jobs directly at the C-level. I think it’s much more important that the skills generally find their way into companies. And that companies take on the topic of AI – including in cross-divisional functions. First, you need to understand the technologies and then establish them in your business model and processes in a meaningful way, but also in your organization and corporate culture.

So we shoudn’t just think of artificial intelligence from a technical perspective?

No, it affects everyone in a company and influences their work far beyond the technical level.

This relationship is not easy. If you enter “AI replaces” in search engines, AutoFill immediately completes “jobs,” then “people.”

It is true. There is a great deal of skepticism and reservations in the context of AI. AI is already bringing about major changes and will continue to do so in the future. It is a pioneering technology. Now, people don’t always like change. That’s why I believe that rejecting AI is a natural and, to some extent, healthy gut reaction. From a risk/reward perspective, many people are even worried.

Do you understand these concerns? Are they justified or based on a myth?

In my view, the statement “AI replaces humans” is the first myth that is often encountered in consulting practice. AI can perform certain tasks better than a human – that is true to a certain extent. This has also been the case throughout the history of technological progress. In the future, it will also affect more fields of activity. I understand the concerns, but humans will not be replaced per se – only activities. On the other hand, people's skillsets will develop in a new direction. New jobs will be created. We have already talked about one example: prompt engineering. The field that this will lead to alone will create jobs that never existed before.

So, new skillset – new mindset?

Yes, it’s now really about asking yourself: How can I do a job with AI that is suitable for me? I also think this mindset is more targeted. Because resisting technological progress is pointless and the question of how I can do a job better than an AI will sooner or later lead to a dead end. AI is here to stay. It’s not going away. We don’t just have to come to terms with it: Ideally, we should draw the maximum benefit from it for ourselves.

What other AI myths do you encounter?

At the level of our project work in SMEs, the following assumptions are certainly true: “AI is too expensive for us,” “AI is more for corporations with the appropriate structures and resources.”

Okay. Mythos Nr. 3?

“We are a medium-sized company. AI is too far removed from what we essentially do here” or “There are no specific use cases for my company.” And I can also understand these assumptions.

Can you dispel them? What do you do to counter these assumptions?

Many models are freely available on an open source basis or can be accessed for a reasonable license fee. There are also very good and very specific use cases in every SME context. For example, companies can consider how they could incorporate artificial intelligence into their customer service processes. Instead of overworked service teams, AI would provide customers with fast, correct and accurate answers. There are also very specific and exciting approaches in the logistics context: Inventory levels and route planning can be optimized with the help of AI. I can evaluate and manage product ranges from a much more comprehensive perspective and, in particular, integrate pricing dynamics into my business processes that were previously unthinkable. The list of possible use cases is very long.

Which current AI developments make you uneasy – regardless of the direction?

That’s a long list too. I’ll limit myself to two topics. Firstly, generative AI. It has made a quantum leap into the mainstream in the last 18 months. As a data analyst, I ask myself in the field of my work how I will communicate with AI in the future. Specifically, how can I communicate with a dataset in natural language without necessarily having to descend to the level of coding or formulaics? I can hardly wait for Microsoft to integrate “Copilot” into the Office world as standard and for this interaction to pick up speed. Which parallel developments make me anxious: Definitely AI and weapons systems – it’s already a reality and the research is going on. People are rightly talking about a red line here. That worries me.

There are numerous conflicts of values in the development and use of AI. What trends have you observed over the last ten years when you focus on different generations and how they deal with AI?

I have observed that it is more natural for each subsequent generation to regard new technologies as something completely normal. The younger generations use new technologies a lot and very quickly. At the same time, I notice that the skepticism we talked about is far less present among the younger generation. As far as security aspects are concerned, this is a question that older generations are increasingly asking. After all, they have grown up with a different set of values and different ways of living and working. On the other hand, it opens up significant possibilities in terms of security. The question is: how do we ensure that we can control this technology – in particular, that we can guarantee the security of the data that is processed in it without letting these considerations slow us down too much?

What other questions do you have about AI in an advisory context?

What concerns me is where the journey is going. Because it’s clear that AI is here to stay. But how do we use it to develop products? How do we create concrete solutions from it? Solutions that enable us to use AI safely and generate real added value for us and our customers? That’s what it’s all about.

Thank you very much for the interview, Mr. Trapp.

 

What questions do you have about AI? Get in touch with us! We look forward to hearing from you.

Get industry insights now!