Artificial Intelligence (AI).

The term is almost 70 years old; it was coined at a conference in Dartmouth in 1956. For a long time it mainly meant rule-based systems that experts programmed by hand. Today's surge comes from machine learning: instead of being given rules, a system learns from a very large number of examples which patterns belong together. A special form of this are large language models such as GPT, Claude or Gemini, which have learned to understand and generate language from vast amounts of text.

In everyday use, three types can be distinguished. Recognising AI classifies — spam in the inbox, faces in photos or fraud patterns in payments. Predictive AI estimates what is likely to happen, such as demand or default risks. And generative AI creates something new: texts, images, code or summaries. Since 2022, public discussion has revolved almost exclusively around generative AI, because for the first time it became usable by anyone without specialist knowledge.

Important for use in companies: AI does not understand in the human sense. A language model calculates which answer most probably fits an input. This works astonishingly well but also leads to so-called hallucinations — convincingly worded, false statements. AI is therefore excellent for drafts, summaries, classification and working with your own documents, but it needs human review for anything that is legally, financially or contractually binding towards customers.

For Swiss SMEs, the greatest benefit usually lies not in spectacular projects but in recurring work: pre-sorting enquiries, preparing quotes from text modules, extracting information from documents, answering internal questions based on your own records. Data protection belongs in from the start, because as soon as personal data flows into an AI tool, the revised Data Protection Act applies.