AI is omnipresent in the discussion, but is rarely really understood. Headlines about revolutionary technologies and disruptive business models create a mixture of enthusiasm and uncertainty. However, in order for a company or institution to make well-founded decisions, it does not need a vague sense of the future, but concrete, action-guiding insights.
Artificial intelligence: a paradoxical phenomenon
Artificial intelligence (AI) is a phenomenon full of ambivalence: it is either dismissed as temporary hype, praised as a revolutionary economic paradigm or underestimated in terms of its influence - but is always accompanied by a considerable degree of ambiguity. This ambivalence is no coincidence, but the result of profound geopolitical shifts, continued rapid technological progress and a regulatory framework that is still in the process of being created. One thing is certain: the transformation triggered by AI in the wake of the ongoing digitalization of large parts of our working lives is already having far-reaching economic and social consequences. This is because AI means both the possibility and the necessity of increasing productivity and innovation in almost all sectors - a trend that, as we at the AI Alliance experience in our daily work, is also continuing for SMEs in particular.
While companies and societies are discussing the potential of AI to create new products, services and experiences and are increasingly able to exploit it for themselves, the long-term effects raise pressing questions. What answers can AI provide to the most critical challenges of our time? How can technological progress and ethical responsibility be brought into a sensible balance? And above all: how can "real" and long-term added value be created through and with AI in ongoing operations?
It is now a matter of strategically utilizing their potential to shape a future that benefits as many people as possible. The course we set today - whether in the form of small or large investments, overarching regulations or individual entrepreneurial decisions on the ground - will shape the development of our society, industries and realities of life for decades to come.
This is precisely where this series of articles comes in. Our aim is to create points of contact and open up a well-founded dialog on the strategic use of AI. Engaging with this technology is not an optional activity for those with an affinity for technology, but a business and social necessity. It requires smart, reflective and forward-looking decisions. This requires insight, courage, partnerships, a high degree of tolerance for ambiguity - and, above all, knowledge.
This first article in the monthly "AI Insights" series lays the foundation with the question:
What do we mean when we talk about "artificial intelligence"?
New series of articles from the AI Alliance
The rapid development of artificial intelligence (AI) is challenging us all - especially in small and medium-sized enterprises (SMEs) and municipal institutions.
In this series of articles, we would like to shed light on and discuss essential questions relating to artificial intelligence - from the meaning of individual technical terms, ethical and moral issues or the interaction between humans and machines, to the classification of global developments at regional level and the management of regulatory challenges - always with a clear focus on the fields of application in business and society.
About the author: Dr. Jan Zipp is responsible for AI Strategy & Innovation and Community Management at the KI-Allianz in Tübingen. He has been working in the AI and deep tech sector in Germany and abroad for over ten years. His focus and publications are in the field of institutional and artificial (computational) creativity research, human-machine interaction and the impact of AI on the future of human work.
Why a definition alone does not help
The term "artificial intelligence" is initially a shell that can be filled with a variety of different meanings. Although almost every person probably understands the term, they interpret it very individually. At the same time, we are not always aware of whether our counterpart follows our interpretation or uses a completely different, subjective understanding of AI. AI as a term is therefore multi-layered, rarely clearly defined and often misunderstood.
It is therefore important to first distinguish between AI as a scientific concept and AI as a technological application:
- Artificial intelligence (AI) in the general sense describes the scientific field of research that deals with the development of intelligent systems.
- AI systems, on the other hand, are specific technical applications that simulate and automate cognitive abilities such as pattern recognition, prediction or language processing. The OECD, which is constantly adapting its definition to current circumstances and is therefore also mentioned here from the multitude of offers, defines AI systems as machine-based systems that perceive their environment by collecting data, interpreting it and learning from it in order to adapt flexibly to new conditions and achieve certain goals (OECD, 2024).
There is a special class within these systems: AI agents. These are characterized by the fact that they not only process information, but can also react autonomously to their environment, make decisions and adapt their behaviour. Intelligent assistance systems, adaptive recommendation systems and autonomous vehicles are examples of AI agents that go beyond pure pattern recognition and act independently.
A differentiation of the terms is often ignored in practice - in most cases, "AI" actually refers to an AI system or an AI agent. Implicitly, the question "What is AI?" does not only deal with what AI is, but also how it can be used in a meaningful way.
Nevertheless, the question of whether a technology can actually be described as "artificial intelligence" or whether it is more a matter of classic algorithms or machine learning is far more than an academic quibble. This debate forms the theoretical foundation for our understanding of (artificial as well as human) intelligence and automation - and therefore also has far-reaching practical implications. It forces us to question our own relationship to technology, holds up a mirror to our human actions and helps us to critically reflect on the limits and possibilities of the role of algorithmic systems in business and society. So how can AI be defined more clearly?
Not all AI is the same
In popular media portrayals, AI is often outlined as a machine that has a human-like understanding of the world and similar cognitive abilities - not only solving specific tasks, but possibly even developing its own consciousness. However, such a system only exists as a theoretical concept so far and is referred to (in very simplified terms) as "Strong AI". In contrast, there is "Weak AI" - specialized systems that have been developed for clearly defined tasks, such as image recognition, language processing or machine translation. They are often highly efficient, but remain strictly limited to their respective functional area.
This distinction between strong and weak AI serves as an initial orientation, but falls short of accurately describing the current state of the technology. Rather, modern AI systems operate in a spectrum between these categories. Research is differentiating increasingly finely in order to better reflect the actual state of development of AI:
- Narrow AI: Systems that solve specific tasks with high efficiency - for example in image recognition, chatbots or recommendation systems. This form currently dominates the market and forms the basis of almost all practical applications.
- Broad AI: Systems that can process knowledge from multiple domains without having true generalizing intelligence. Examples include multimodal models such as GPT-4 or DeepMind Gemini, which combine language, images and logical problems.
- Artificial General Intelligence (AGI): A hypothetical concept for an AI that can learn flexibly across different domains and adapt to new contexts - similar to human intelligence.
- Superintelligent AI (ASI, Artificial Superintelligence): A hitherto purely theoretical level that describes an intelligence that would be superior to human intelligence in all respects and will very probably - despite all the technological quantum leaps - remain a dream of the future (some even say: must remain ! From both a logical and an ethical point of view).
This classification shows that AI is not a monolithic concept, but a technological continuum that is becoming increasingly differentiated as research progresses. While the public debate is often oriented towards visionary future scenarios, actual progress to date has largely been in the area ofnarrow and increasinglybroad AI models.
However, this classification is not exhaustive either. The rapid development of AI technologies requires a continuous adaptation of definitional approaches, as new paradigms emerge that do not fit seamlessly into existing categories. One example of this is the often-discussed unsupervised learning, which can be located between narrow and broad AI. It describes algorithms that independently recognize patterns in data - without explicit rules or predefined labels. This ability to autonomously recognize patterns and extract knowledge pushes the boundaries of what AI systems can do, but remains limited to specialized fields of application.
Even this brief description of some key concepts shows that there is no "one" AI, but rather a multitude of specialized technologies with different degrees of autonomy and areas of application. While some AI systems serve as tools for data analysis and automation, AI agents are increasingly taking on adaptive and interactive tasks. The challenge lies in finding the right solution for the respective context and tapping into its potential in a targeted manner - despite or perhaps because of all the uncertainty that artificial intelligence will continue to bring in the future.
Asking ourselves what is actually meant in each specific case when we talk about AI seems to be a sensible first step. We will discuss what further steps need to be taken from our point of view and how the path can be successful in further contributions. We look forward to exchanging ideas on this - whether digitally or in direct discussions on site!
How the AI Alliance can provide support
We at the KI-Allianz Baden-Württemberg see ourselves as a door opener for possible solutions and companions on the path to the smart use of artificial intelligence. With our network and extensive knowledge, we support interested parties as trailblazers, listen to them, enter into discussions with them and open up new perspectives.
The KI-Allianz Baden-Württemberg sees itself as a door opener for AI solutions and as a companion on the path to the intelligent use of artificial intelligence. Our network combines expertise and knowledge to support companies and institutions. From process optimization to new business models, we point out options and open up new perspectives - with or without AI.
Would you like to find out more about artificial intelligence and the Baden-Württemberg AI Alliance?
Then subscribe to our newsletter!
