on
Artificial intelligence is the next big leap in digital evolution
on
Artificial intelligence is the next big leap in digital evolution. Anyone who doesn't engage with it will soon look like a horse-drawn carriage next to a racing car. There is no doubt about that for Boris Eckstein. In an expert interview with our editor Franziska Pleßke, the experienced interim manager for digital innovation projects discusses the beginnings of AI and shows how companies can use artificial intelligence profitably with the help of an MVP approach including a build-measure-learn cycle - practical examples included.
Mr. Eckstein, since the launch of ChatGPT, everyone has been talking about artificial intelligence and the AI hype is almost omnipresent. Yet artificial intelligence and AI applications are not as new as many might think. Is it all just old wine in new bottles?
Boris Eckstein: It is surprising how far back the beginnings of artificial intelligence go. There were already fascinating approaches more than 100 years ago - long before the first computer was invented. Google "El Ajedrecista", for example, which means "the chess player", an electromechanical machine that could play rudimentary chess. Artificial neural networks were described as early as the 1940s, the famous Turing test dates back to the 1950s and ELIZA, the pioneer of chatbots par excellence, saw the light of day in 1966.
The topic of artificial intelligence has evolved over many decades. However, it is only now that we have the necessary technical requirements to make AI powerful enough. Something like large language models (LLMs), which are based on artificial neural networks and vector databases and are trained with high-performance processors and the content of millions of websites - in terms of function and performance, this is a huge difference to the chatbots or virtual language assistants we saw just a few years ago. In this respect, GenAI technologies are definitely ushering in a new chapter in the field of artificial intelligence.
New AI solutions are springing up at ever shorter intervals and are developing rapidly. Why should, no, must companies address the topic of artificial intelligence?
BE: Quite simply: there is no way around it. Look back 20 or 25 years. Back then, companies were asking themselves why they should go online and why they needed a website. Today, it is unimaginable not to have digitized all essential business processes. Even companies that don't create digital products or services or don't do most of their sales online see themselves as software companies to some extent. Just think of supermarket chains or car manufacturers.
Artificial intelligence is the next logical evolutionary step, no, evolutionary leap. It will increase digital performance enormously, drive process automation, change existing business models and create new ones. Generative AI is not coming, it is already here. And it is being used productively with enormous success - in B2B and B2C alike. The "Amazon CodeWhisperer" programs in the AWS cloud, AI-based image processing is available to everyone in the new Google smartphones and the "Copilot" in Microsoft's Bing search provides search results summarized in natural language. Anyone who doesn't keep up with this development will soon look like a horse-drawn carriage next to a racing car.
Let's get specific. How can companies use AI profitably and how should they go about it?
BE: What I'm about to say may sound banal. But in practice, it's often not that simple. With or without AI, the main thing is to develop a functioning business case. This includes the following three aspects:
- It must create business value.
- It must generate user value.
- It must be feasible - especially technically, but also organizationally.
The way to get there - and this is not new either - is via a so-called Minimal Viable Product (MVP), which in turn should be integrated into a Build-Measure-Learn-Cycle. An MVP is a first functional iteration of a product. The purpose of this is to check the business case and learn from it as quickly as possible. In user experience, they say "fail often, fail early". of course, "to fail" here does not mean to fail, but to learn, to recognize errors as early as possible and thus to develop an ever better product. Especially with new technologies, this kind of approach is worth its weight in gold.
Can you give us examples from your practice where you have successfully implemented AI applications? And what benefits have resulted for the companies?
BE: I established a chatbot for a German cable network operator. The aim was to automate customer center services. This is a good example of the MVP approach described above: we identified use cases that bring an economic benefit, have a customer benefit and are easy to implement. The first use case was quickly found: Disruptions. If an Internet node goes down, the number of calls to the customer center skyrockets. All capacities are tied up at once. Customers are then kept on hold for a long time and the well-trained call center agents have the same undemanding conversation a hundred times in a row - "I see, you have no Internet. What is your address, please? I'm sorry, we have a problem. May I send you a text message when the fault has been rectified? Sorry, have a nice day." The advantage here was that the company had an app in use that was already being actively used by customers. This allowed us to use the chatbot seamlessly in an established channel.
I also had the opportunity to develop the user experience in an AI-supported identification process for a German shipping service provider. A wide variety of AI technologies were used here, including OCR (Optical Character Recognition), facial recognition to check whether the person in front of the camera is identical to the ID photo and a "liveness check" that recognizes whether there is actually a person in front of the camera and not just a photo. A key challenge here was to design the new AI tools, which were used as assistance systems for the call center agents, in such a way that they accelerate processes and ensure greater security. As a UX expert, I can assure you that the interaction between humans and AI is not only very exciting, but also absolutely business-critical.
At the beginning of the year, I also published a seminar myself: "IT know-how for specialist managers". With ChatGPT, I was able to create almost the entire glossary, among other things. This reduced the editorial workload enormously. I implemented some lessons with a video avatar. This does not yet have the quality of a human presenter. But it is acceptable and it reduces the effort required for video production to a fraction.
Companies often simply lack the capacity to tackle such future-oriented projects. However, if introduced and implemented correctly, AI solutions can free up resources and increase competitiveness. How can interim managers provide support here? And what are the advantages compared to internal implementation?
BE: It's not just about capacity. It's about making structural changes in a company. Such projects are often easier to implement with external parties. Firstly, they are not dependent on an existing position and secondly, they do not have any competing career intentions. What's more, companies often need to develop processes and skills first. AI applications, especially GenAI technologies, involve entire process chains. Data needs to be provided, models trained and updated. At the same time, permanent quality control must be introduced, security measures must be established and the user experience must be managed. And so on.
Interim managers can help here by bringing technical and structural knowledge. They identify relevant business cases, build structures and establish processes. In short: interim professionals get things up and running and then bring them into line. And then they can go, because their job is done. And to be honest: with interim managers you also reduce the entrepreneurial risk, i.e. the danger of a bad investment. Because in the event of failure, external employees are easy to dismiss - quite the opposite of permanent employees.
Honest words! Finally, is there anything else you would like to say to companies that are just starting to get to grips with the topic of AI?
BE: My appeal to companies: Try out new things and show an honest willingness to learn! Look for simple use cases for AI, get them off the ground and see where the potential lies. If things don't work the way you thought they would, you haven't failed. You've only failed if you don't learn anything from it.
Mr. Eckstein, thank you for the interesting conversation and the personal insights!
Would you like to drive forward a transformation project - in the field of AI or beyond? We have the right solution for you. Feel free to contact us at any time!
