Retrology instead of disruption? Why AI euphoria is often backward-looking
Preface
As someone who has been working with AI systems since the early 1990s, called expert systems back than, and considers technology to be increasingly important in the context of project missions at iMB.Solutions, I follow developments in the field of artificial intelligence with great interest. In particular, automation, IoT, digitalization, machine learning, generative AI, and big data analysis are playing an increasingly important role. When it comes to implementing specific project requirements, small language model (SLM) applications are relevant in virtually all cases.
Time Line
My engineering education in hydrodynamics and factory plant planning brought me into contact with complex, chaotic, and non-linear natural sciences at an early age. Even back then, in the 1990s, we used expert systems to simulate flow behavior. Although these systems ran on gigantic workstations and were slow, their philosophy is not dissimilar to that of today's generative AI. My thesis at Mercedes-Benz dealt with the development of a workstation menu and simulation program based on a 3D volume modeler for production plant planning. It was already about virtual space in production, the analysis of data and material flows – parameters that were attached to 3D models to simulate processes.
Intergraph CAD Workstation
source: https://forums.overclockers.co.uk/threads/old-computer-and-pc-brands.18997982/page-4
My postgraduate thesis in international business also focused on the automotive industry, more specifically on the development of a dynamic route optimization system for logistics in Europe. These experiences have repeatedly brought me into contact with AI - expert system - applications. In recent years, I have also completed advanced training as an Advanced Digital Engineer at the HassoPlattner Institute.
Today’s Project Realities
At iMB.Solutions, the use of technology in our project missions has steadily increased. Since Q3 2022, we have been using generative AI models in customer applications, mainly in the areas of process mining, supply chain optimization, strategic scenario generation, and design thinking in the development of new business models and transformation projects. In late 2023, the idea arose to develop a specific B2B business model with massive use of LLMs and SLMs. This has been extremely successful, and we are about to launch the application on the market, with a spin-off as an independent start-up very likely after a test phase.
Recent Experiences
I recently attended an HR B2B service provider event on AI in HR in São Paulo. The list of speakers was promising, with experts who had both long-standing experience with AI applications and were startup founders. However, I could hardly believe what unfolded there. It was a mixture of a seminar on philosophy and social sciences and the mystification of generative AI.
Questionable approaches and a lack of understanding - Retrology rising!
A startup founder presented a system in which physical car salespeople receive a trained generative AI bot on their mobile devices to improve their sales pitches. The success of these salespeople is said to be 20-30% higher than with the traditional method. But the question is:
Where is generative AI being used intelligently and with an eye to the future here?
In my view, what prevails here is “retrology” – a reference to “futurology.”
This use case supports car salespeople in their arguments with a generative AI bot. This is a throwback to the era of Manchester capitalism, where the sole focus is on optimizing people. You need an empathetic but uneducated car salesperson, because AI is supposed to take care of the rest. This reduces people to a cost factor. Yet generative AI has the power to free us from monotonous work and bring us back to what we are really good at: creativity. The case of the car salesperson looks back to a distant past, a time we wanted to leave behind. New technology is being used to lower the demands on humans while still optimizing their productivity. But that is precisely not the point of generative AI. It should liberate us, not regulate us. Creativity, strategic thinking, humanity – that is our domain. Not that of machines.
A retrology, caught in retrovision, from a startup founder. Can’t believe it.
Another case was also disturbing. A professional argued that pilot projects for the use of generative AI should only be carried out if clear KPIs for ROI are defined. This suggests that a supposedly experienced project manager demands quantitative goals and metrics from the outset, otherwise one should not even begin a pilot project to implement generative AI in the company. However, my experience with pilot projects over the last 30 years is quite different. One must start with qualitative targets to form a compass and then define quantitative goals such as KPIs and ROI in further steps. The professional who wants to set quantitative targets in a completely new field of application right from the start seems to be more “in love with Excel.” Reality? Successful pilot projects always start with qualitative goals. Only with growing understanding and iterative loops are quantitative KPIs defined.
Anyone who skips this step is not planning innovation, but a controlling project.
A third strange point about the event was that only LLMs were discussed. There was not even a mention of SLMs (small language models). Especially in real business applications, LLMs are not the determining factor, but rather the systemic adaptation and construction of SLMs for business use. It is difficult to say why the professionals at the event were operating at this level. It seems to be a mixture of ignorance, uncertainty, and mystification. Yet SLMs are the operational lever in medium-sized businesses, in HR process automation, in support, in planning, in reporting. LLMs are good for prototyping and exploration – but in operation?
What counts there is efficiency, adaptability, data security. SLMs deliver.
AI needs vision, not nostalgia
One thing is clear, however: it was really weak. What remains? An uneasy feeling. The industry seems torn between mystification, superficial knowledge, and business rhetoric. What is missing is a clear stance, deep understanding, and—yes—a certain courage to be honest.
At iMB.Solutions, we don't believe in AI as an end in itself. We believe in technological empowerment, in new business models, in real transformation. And sometimes that means daring to disagree when everyone else is nodding in agreement.
Because the future doesn't belong to those who let AI look backward—it belongs to those who let it think forward.
