We write a lot of text. Much of what we do has its origins in text. Regardless of whether it is a script for a film, planning for a distance education or parts of an e-learning.

Here we usually use ChatGPT or Microsoft Copilot to get suggestions for content, outline or tonality. When it comes to facts and pedagogy, however, it is more manual work. We want to be sure that what we say is correct and that it is as educational as it needs to be. We want the last word, so to speak.

Images and illustrations generated by AI work best as they do not need to be so specific, in much the same way as the large image banks. If there are specific images of, for example, a person in a certain professional role, a situation at a workplace or the like, we still traditionally work with real photographs or illustrations produced by ourselves.

However, we can use AI to quickly and easily re-edit an image, remove a background or add an element.

In our industry, synthetic voices have long been an alternative to the studio-recorded voice. With AI-powered TTS (Text-to-speech) services, we've got another option. Here it is important to keep track of how what is to be said sounds when it is not said by a real person. An AI can easily stumble on abbreviations or have problems with professional jargon. If the AI service is given reasonable conditions, it is, however, a cost-effective alternative to the studio recording. However, we do not clone voices other than our own.