How AI is making progress in media production

by: Luis Rodríguez Lombardero, Strategic Sales Manager at Innova-tsn.

In the world of media, like in many others, there is a great interest in analysing the capabilities of AI to improve business processes, but the potential threat it may pose to the labour structure, both in more ardous tasks and in creative activities, is also considered. In this series of articles, we aim to analyse and discuss three aspects of the power of AI to influence the future of creation in the media:

Opportunities for a future-oriented editorial system within journalism, where we will consider both organizational and technological transformations.
Possibilities of AI image analysis in the age of short videos.
Use of machine learning to extract the dominant emotion from a video segment and generate appropriate background music for it.

First, we will delve into the future of news journalism with AI, trying to predict what an AI-powered newsroom would look like and the organisational changes this would entail. In doing so, we will draw on our experience and a study conducted for several Dutch media outlets. How did they feel about the results and how much influence can they really have on the direction of AI? Secondly, we will discuss an advanced Japanese AI technology which seeks to benefit those with busy lives: a system able to edit a TV programme with an arbitrarily short duration. We will see how neural networks classify and edit the most noticeable video segments and how these summarised videos are already being tested nationwide. Finally, we will look at an extraordinary artificial intelligence technology that can identify the nature and intensity of emotion in a dramatic scene and then compose and perform background music to accompany it.

Editorial systems in the media and organisational changes driven by AI

Editorial systems are key in structuring the information for readers and in managing contents and communication with users. Their usability, flexibility and integration ability are critical in digital journalism. Journalists use these systems every day to process articles communicate and archive content. However, these systems are often old and not fully adapted to technological advancements such as AI, which offers opportunities for their revitalisation. The new hybrid and digital ways of working demand a redesign of editorial systems which involves both knowledge of software engineering and artificial intelligence, not forgetting a deep understanding of the journalistic business. It is crucial that journalists participate in the design of these systems to encourage collaboration and creativity, rather than being mere users of a system defined by technology alone. The COVID-19 pandemic revealed shortcomings in editorial systems, which were not prepared to fully support hybrid work. The rapid advance of AI poses additional challenges, as editorial systems must adapt to incorporate AI tools to help with tasks such as transcription, translation or image editing. Our analysis seeks to define what organisational, technical and functional changes are needed in editorial systems to respond effectively to the needs of modern journalism, geared towards a practice that is more integrated with AI and suitable for hybrid work.

Main conclusions on editorial systems

Editorial systems need to adapt better to today’s journalistic practices, integrating common tools such as search engines, social media and productivity software such as Excel, Photoshop, Word, Teams, etc Simplicity and ease of use are essential, as overly complex systems can be distracting and complicate the work. Journalists often face tedious processes for entering data required for publishing articles, such as title, subtitle, main text, image and author, which can lead to errors and SEO problems due to excessive time spent on metadata and archiving. Maximising automation is recommended to improve the search and maintenance of articles. In addition, there are significant opportunities to improve multi-channel publishing, allowing content to be automatically adapted to the specifications of different platforms from a neutral text. Finally, it is crucial to improve communication and collaboration within sistemas editoriales, are currently not sufficiently equipped to support these functions effectively. For this reason, they often rely on external tools for planning and collaboration. Enabling simultaneous work on the same file within the editorial system is essential to optimise these processes.

AI tools and applications in the media

Numerous experiments with artificial intelligence (AI) are being carried out in the Media sector to discover, research and verify stories, although the integration of these tools in editorial systems is still limited. Collaborative strategies have been developed with newsrooms to design an AI-driven editorial system, identifying necessary functionalities and their impact on ways of working and journalistic practice.

In an interactive workshop where AI-driven tools were visualised and, based on previous interviews and our AI inventory, eight AI tools were designed:

Tools that can be easily integrated from an organisational and technical point of view:

Smart filing: It automates the filing of metadata, thus reducing a journalist’s manual workload.
Integration of experts: It facilities collaboration between multiple expertes in the reation of stories, supporting diversity.
Sentiment analysis in social media: It enables journalist to quickly get an overview of ocurrent sentiment.

Tools that are technically accessible but organisationally challenging:

Smart planner: It schedules appointments automatically, considering personal and team schedules, requiring employees to share their calendars.
Responsive text editor: Able to create articles form previous texts, similar to tools such as ChatGPT. This means the need for clear policies for its ethical use.

Tools with an emerging technological development:

Digital twin: It allows virtual presence on TV programmes with no need of being physically present.

Future-oriented tools, with technical and organisational challenges:

Smart drones:They automate the recording of events.
Hybrid newsroom: Operable from any location.

These tools are designed for enhancing efficiency and adaptability of journalistic processes, although they require organisational and technical adjustments for their complete implementation.

Main conclusions on the implementation of AI in editorial systems

Experts regard the smart filing system and the smart planner as key and practical tools for future information processes.

Journalists see AI-based text tools such as responsive text editors and sentiment analysis as promising, although they pose organizational challenges requiring setting clear controls and a greater understanding of AI for its effective use.

Although the smart drone camera and sentiment analysis tool offer technical innovations, there are perceived limitations to their practical value due to operational difficulties and privacy concerns. Interpretation of social data can also be problematic due to possible under-representation of certain groups.

In short, it is expected that future editorial systems include AI to simplify tedious task such as content editing for multiple platforms and file organisation. However, the integration of automated writing tools requires a deep understanding of AI and strong supervision mechanisms to ensure a responsible implementation.

It is crucial that editorial systems are aligned with journalistic practices and can adapt to future organisational developments, facilitating collaborative working and multidimensional storytelling. The pandemic has highlighted the need for systems that support both face-to-face and remote working, and it is hoped that future multidisciplinary research will delve deeper into these areas to enhance editorial systems with AI technology.