Since the advent of AI, major outlets have tried to incorporate it into their newsroom workflow. From Sports Illustrated’s use of AI to generate fitness news to Yle’s using AI to translate Finnish news into Ukrainian for Ukrainian citizens, AI has undoubtedly entered newsrooms, yet many journalists don’t understand its full picture.

This is the black box problem, where we lack the skills to discern how AI tools work. This generates its own set of problems, such as lack of transparency, plagiarism and copyright infringement.

Despite the risk of not understanding AI in its totality, many journalists show enthusiasm for using it. A study of various professionals in Denmark found that journalists were the second highest adopters of AI, with 64% using it in some capacity.

AI and its integration into journalism was a big talking point at the 10th European Communication Conference (ECREA). Last month, researchers from European universities presented their research on the impact of AI and automation on journalism and journalists. The bulk of studies focused on audiences’ and journalists’ perceptions of AI and concerns over the ethical usage of AI.

The Fix has compiled three key findings that offer valuable insights from the latest research, along with practical ways news outlets are putting these strategies into action.

Lack of knowledge among journalists

As mentioned above, many journalists lack a clear understanding of how AI functions and often struggle to distinguish between what does and doesn’t qualify as AI usage. This confusion is further aggravated by a lack of clear guidelines at the organisational and national / international levels.

Researchers from the Fontys Academy of Journalism, Netherlands conducted interviews with journalists from regional Dutch newsrooms on their view of AI and its integration. Many journalists were uncertain about their ability to deal with AI and this hindered their ability to use the available resources. AI illiteracy also hampered the journalists’ ability to communicate their needs to the IT professionals.

Lack of time was cited as one of the key reasons why journalists were unable to upskill. Here, are two resources that can give journalists an easy start to learning AI in a short time:

The Fix AI Course is a tailor-made guide for news professionals to learn and implement AI. In 8 weekly instalments as newsletters, readers get tips on AI usage by The Fix contributor and Slow News editor-in-chief Alberto Puliafito.
Another free resource on AI and journalism is by JournalismAI. After completion of this 6-module course, participants can even receive a certificate of completion.

Some audiences are willing to pay for AI-generated news

University of Zurich researchers, Dr Daniel Vogler and team, found that audiences are not willing to pay for news entirely made using AI – except for younger and less educated individuals who look at AI more positively, hence are willing to consume and pay for AI-generated news.

This is evident in The Economist’s Espresso app which is designed for younger readers. Last month, they launched an in-app translation feature powered by AI. With AI translations they aim to reach a wider audience speaking Mandarin, Spanish, French and German.

Reiter, Hess and Adolf from FHWien der WKW, Austria found that in some cases GenAI usage in news is seen as more trustworthy than human-made news. Their reasoning – journalists may carry bias but AI is impartial. They felt that the role of the journalist was to provide “quality control.” Additionally, AI usage for editing and applying filters for visual media was acceptable, especially for sports coverage.

Argentinian local news outlet Diario Huarpe has been using automation to churn out multiple sports stories despite having only onr reporter cover this beat during the weekends. Their AI bot prepares the article by feeding on the result of the game and the structure of the article. This helps the reporter to focus on interviews and editing the article.
Automation in sports is also seen with ESPN, a major American sports broadcaster, where AI analytics is used to provide a better viewing experience for the audience by providing them with game insights, player information and updating them with real-time analyses and predictions.

News personalisation for better audience retention

Researchers from Germany’s University of Mainz and Ludwig-Maximilians-Universität München found that the acceptance of news recommender systems (NRS) for sending news of diverse viewpoints is higher only amongst those who already consume diverse news. News publishers should hence pour more effort into personalising the content to the reader for better results.

Norway’s oldest newspaper Adresseavisen, has used AI to personalise the homepage for its readers. However, they found the best results when personalisation strategies were used for suggesting the next read at the bottom of the article. This helped Adresseavisen to make sure that the readers stayed longer to consume more and even get them to subscribe.
Using AI for personalisation is not only restricted to webpages but even to newsletters. The Times used their AI tool JAMES (Journey Automated Messaging for Higher Engagement), to learn the way their readers preferred their newsletters. Implementing the findings of JAMES helped them reduce subscriber churn by 50%

A common thread in most of this research is the emphasis that this technology is at a stage where complete automation is not possible. A human in the loop should be the norm to eliminate major problems. AI is for people, not a way to get on without them.

Research studies referenced in this article

Source of the cover photo: Ahmad Dirini via Unsplash

AI and automation in journalism – latest insights from research

Priyal Shah

Here are the latest findings of using AI in journalism alongside practical examples from the industry

Author

Priyal Shah

‌

‌‌‌

‌‌‌