+41 544 36 40
  • en
  • de
  • Text revolution through AI

    Our goal is to revolutionise texts AI

    Artificial intelligence (AI) – an ever-growing topic for all areas of society, but especially for the media and content industry. However, AI trends such as image generation or individual film suggestions are just the beginning. AI has the potential to revolutionise the everyday work of editorial offices – with automated text generation and processing. It has enormous potential, as pioneers in this field already show.

    “AI trends will continue to dominate the media world in 2023. It discusses how natural language processing makes the emotional impact of content understandable to users, how it evaluates digital usage data for recommendation systems and personalised content, or how it creates fictional content, storytelling and images. One topic receives less attention, however – even though  it can have the most significant impact on the way we work in media: automated text generation and processing by AI assistant tools,” says Michael Keusgen, CEO of the media tech company Ella Media AG. It has enormous potential to transform the media and content industry and to efficiently adapt to changing requirements.


    AI as an editorial assistant

    Times are challenging: editors are under a lot of pressure to perform due to a lack of resources. In addition, publishers want to cover ever more topics and formats. In an economically strained situation, they have to improve their reach and thus monetisation.

    AI can now not only support news journalism and content production with translations or subtitling – but also directly write texts. “By generating high-quality content that search engines rank, AI-based assistance tools relieve newsrooms of time-consuming routine work. When AI creates its own short news items from agency sources, it increases coverage,” explains Keusgen, who himself was active as a film and television producer in the media, sports and entertainment industries for many years.


    The text (r)evolution

    Swiss-based start-up Ella has long recognised the potential of AI in text creation and processing. Its AI-based language models and software products generate high-quality content for news and media. The News Assistant Re.lease – an AI-based copywriting platform – supports editors by automatically converting existing news texts from agency sources into their own texts, thus preventing duplicate content – simply, efficiently and factually correct.

    Editors do not have to fear for their jobs: they select relevant topics on the platform at the beginning and take over the final editing at the end. Re.lease reduces the time spent on this routine work by up to 50 percent. Media companies can thus publish with fewer resources more unique content, with which they can improve their reach and search engine rankings.


    Responsible AI for high-quality texts

    “AI is a crucial tool in ever more industries – including the media and entertainment sector. Here, we are doing pioneering work with our quality-focused AI models developed specifically for the German language,” explains Keusgen. The quality of the automatically generated texts is a top priority for Ella. The company lives up to the responsibility associated with AI-based content creation: Ella has developed the first fact-checking model for machine-generated German texts. It checks whether the factual situation of the machine-generated text variant is identical to that of the source text – even if the facts in the original text are only implicitly derived from the context. In addition, Ella improves its products through the use of large, meaningful and high-quality training data sets, as well as extensive human feedback through an internal proofreading team. During the development phase of new models or updates, Ella also employs test readers who conduct intensive text quality analyses using linguistic-empirical methods. “We attach great importance to the constant exchange with our quality assurance. Together we work out what a good and factually correct text is and how this fact can be measured. That’s the only way we can better identify weak points in the models and enrich training data in a more meaningful way, which increases the quality of the texts “, says Keusgen.

    Ella thus promotes high-quality journalism that can counteract news fatigue: according to the Reuters Institute Digital News Report, the proportion of people worldwide who often or sometimes avoid news is growing. News fatigue is the result of, among other things, declining trust in media, and is also partly caused by fake news. News assistant tools like Re.lease thus complement special AI tools that are already able to automatically check facts in texts and identify fake news.

    “However, our vision does not only encompass the news media: in the next development steps, we will also launch products that support journalists and content creators as fully-fledged co-creation assistants to turn their ideas and research into creative articles. After that, we want to focus on multimedia and entertainment content,” says Keusgen, looking to the future.


    Press contact:

    Fink & Fuchs AG
    Patrick Rothwell
    phone: 0611 -74131 – 16