Project und Key Account Manager, Retresco
ChatGPT, GPT-4, Google Bard or Data-to-Text – generative Artificial Intelligence (AI) brings disruption to content creation. It supports the generation of idea sketches, channel and target group-specific content variations, multimedia representations and much more, by being trained on extensive data sets. These models have the potential to sustainably change or even revolutionize the preparation and consumption of news. Generative AI will have a significant impact on the way we communicate and work specifically in and for publishing in the future.
However, this development is still in its infancy. The central questions in this context are: How can generative AI be sensibly applied in the media environment? What added value does the technology offer? How can publishers and content creators successfully utilise and master the technology? What influence does AI have on business models and their sustainability? What challenges and risks are there?
Automated text generation has long been commonplace in the media, for example in sports reporting, election results, weather forecasts, traffic news or horoscopes. An extensive basis of structured data plays an essential role here.
Below, we would like to provide inspiration for the possible applications of generative AI, explain possible measures and outline concrete application scenarios, but also address the challenges involved.
Generative AI offers a wide range of possibilities in the media environment to improve quality and efficiency in content creation. While powerful AI systems like ChatGPT already deliver impressive results, they are often still prone to errors, raise legal questions and offer unrefined features. In order not to underestimate the future impact of generative AI on publishing and to shape the industry's development, content, marketing and SEO teams in the media world should deal with the following use cases:
Generative AI systems can automatically create short summaries or teasers for articles to pique users' interest in the full article. For example, an AI-generated teaser for an article on climate change can highlight the main points and spark users' curiosity to learn more.
With the help of generative AI, media companies and content creators can adapt agency contributions to various target groups to develop relevance and informational content precisely for each user group.
AI can identify users' preferred topics in real-time and offer them personalised content recommendations, simultaneously for all visitors and without manual effort. At the same time, generative AI assists in creating personalised news by analysing users' interests and reading habits and selecting relevant articles and topics accordingly. This increases user engagement, dwell time, and reader satisfaction.
Artificial Intelligence can be used to improve texts and articles in terms of search engine optimisation by generating relevant keywords, meta tags, and other SEO-relevant elements and incorporating them into the automatically created text.
Media companies can use AI systems to improve their texts, receiving suggestions for better wording, grammar, and style. This leads to higher text quality and more efficient content workflows.
AI-supported systems also help media and publishers with research by automatically identifying relevant information, articles, and sources on a particular topic. This reduces research effort and allows content managers to focus on analysing and interpreting the information found.
Generative AI can support the production of podcasts and audiobooks from existing texts by automatically generating suitable audio implementations. This enables media and content creators to make their content accessible on new platforms and for additional target groups.
AI can create fitting and meaningful captions for photos and graphics in articles by considering the image's context and the article's content. This increases the impact of the images in the article and conveys additional information succinctly. This type of captioning is already widely applicable.
AI systems can automatically generate suitable illustrations, graphics, or images for articles by analysing the article's content and creating appropriate visual elements. Text-to-image generators such as Stable Diffusion can already place shorter texts into automated illustrations. All these possibilities save time and resources for creating graphics and similar image material.
The use of generative AI in the media sector poses various challenges that should not be underestimated. An increasingly relevant aspect is to reference verifiable information sources. The use of comprehensible and trustworthy sources becomes more critical as automated content generation makes it difficult or even impossible to trace the sources of the information and whether they are factually correct.
Ultimately, the use of generative AI carries the risk of amplifying fake news dissemination. AI systems can generate a large amount of convincing, but false or misleading information in a short time. Distinguishing between true and fabricated information becomes significantly more challenging, potentially harming societal trust in media and institutions. To counteract this, appropriate measures must be taken to combat AI-generated fake news and responsibly and transparently use AI. Human control for fact-checking will play a central role in text creation in the future.
With freely available AI, new questions arise regarding data protection and copyright. Examples are the now commonly spread AI-generated images of famous personalities. The less prominent the depicted individuals and the less known the circumstances, the more challenging it becomes to verify the authenticity of the generated data. Yet, media and publishers must ensure that AI-generated content complies with legal requirements and protects personal data.
One possible solution for media and publishers is automated text generation, combining large language models such as GPT with a data-based approach. Dynamic and static content from generative AI systems like GPT are used within predefined text models to create factually correct and compliance-compliant texts. The text models act as guardrails within which the text generation takes place.
The combination of data-based and generative AI allows for a personalised and channel-specific processing of texts. Tonality, brand presence, target audience appeal, language-specific semantics, and grammar rules are defined and enforced in text models. Once a text model is set up, large volumes of content can be generated and published. An essential advantage of this approach is that human-in-the-loop quality assurance and fact-checking are secured from the outset.
Generative AI opens up new perspectives for media and publishers in content workflows, research tools and digital offerings. In times when alternative options are only a click away, content, marketing, and SEO teams are well-advised to explore AI-supported solutions and identify relevant use cases. A correctly implemented AI system enables higher productivity, automated processes and overall greater efficiency.
However, before media and publishers develop their AI applications, they should ask themselves: Is it worth the effort? Do I have the necessary resources and skills in the company? Or would it make more sense to implement AI projects in collaboration with external experts?
Retresco is a reliable partner for the successful implementation of generative AI in the media environment. Experienced experts from computational linguistics, language science, software development, and project management have successfully conceived and implemented over 250 projects in automation, machine learning and generative AI over the past 15 years.
For questions and further information on generative AI for media and publishers, please do not hesitate to contact us. Talk to us – our experts will be happy to get in touch with you!