Artificial Intelligence (AI)

It is important to note that AI, in today’s practical sense, does not think or understand in the human sense – it recognises patterns in large amounts of data and uses these to solve tasks.

This objectivity is important because the term carries strong connotations. There is a world of difference between the AI seen in science fiction films and the AI that calculates product recommendations or writes texts today. Anyone wishing to use AI effectively in a business setting should focus on the real, clearly defined capabilities of today’s systems, not the myths.

Machine Learning and Deep Learning

Within AI, there is an important hierarchy of terms that is often confused:

The current wave of AI relies heavily on machine learning and, in particular, deep learning. Instead of prescribing every rule to a computer, it is shown many examples and allowed to identify the patterns itself. It is precisely this approach that has enabled breakthroughs in image recognition, natural language processing and generative AI.

Weak and strong AI

A fundamental distinction separates what exists today from what remains the realm of fiction. Weak AI (also known as ‘narrow AI’) specialises in a specific task – a language model, image recognition or a recommendation system. Every AI currently in use is weak AI, even if it seems impressive. Strong AI (also known as ‘AGI’, Artificial General Intelligence) refers to a hypothetical form of AI with general, human-like intelligence across any task. It does not exist and is the subject of research and debate, not practical application. This distinction helps to keep expectations grounded: What delivers value in business is always specialised, weak AI.

Generative AI as the current driver

The biggest breakthrough in recent years has come from generative AI – systems that generate new content: text, images, code and audio. Large language models, such as those behind ChatGPT, are the most prominent example. They have transformed AI from a tool for specialists into something that anyone can use via natural language. For businesses, this has opened up a wide range of new applications – from text generation and customer service to software development.

A concrete example

A medium-sized retailer receives numerous product enquiries, return requests and invoices every day – in text form, unstructured and of varying quality. An AI based on a language model reads these messages, recognises what they are about, assigns them to the correct category and extracts the key information – such as the invoice number or the product in question. What staff used to sort manually is now largely automated, whilst critical cases continue to be reviewed by humans. The AI does not ‘understand’ the issues in the same way as a human, but it recognises patterns reliably enough to save a significant amount of effort. It is precisely here – in recurring, language-based tasks – that the typical, easily measurable benefits of today’s AI lie.

Opportunities and limitations

AI can make businesses faster, more cost-effective and more responsive. But it has clear limitations. AI systems can produce incorrect outputs with great conviction; they are only as good as their training data; and they raise questions regarding data protection, copyright and accountability. Furthermore, with the EU AI Act, there is a growing legal framework regulating certain applications. Responsible use therefore always means: choosing the right use case, verifying the results and leaving responsibility in human hands at the crucial points.

Context

Artificial intelligence is arguably the most far-reaching technology of our time – but it is a tool, not magic. Its value does not stem from the buzzword itself, but from its concrete application to a real-world problem. For small and medium-sized enterprises, the opportunity lies in using AI pragmatically where it delivers measurable benefits, rather than chasing a vague trend. A well-founded, accessible introduction to the topic is provided, for example, by the Federal Office for Information Security (BSI), which also examines AI from a security perspective.