Many in our industry are already using GenAI in interesting ways. The most common theme? Using it as an assistant for busy reporters and editors in relatively low-risk ways, particularly on the back end. For example:
Die Presse uses it to generate interview questions, story ideas, and social media headlines.
Dennik N uses it to predict churn, convert video to text and text to voice and image, and translate stories.
Schibsted uses it to transcribe interviews. After just two months, it has saved journalists more than 3,000 hours of work.
Hearst Newspapers uses it to suggest Web headlines, SEO keywords, and URLs.
Aftonbladet uses it to assist journalists, with a tool that proofreads and gives feedback on sentence structure, finds repetitions, and weaknesses in reasoning.
News Corp uses AI to help it produce 3,000 Australian local news stories a week.
The Daily Maverick uses it to generate summaries and will experiment with headline writing as well.
Bloomberg has built its own LLM that can provide suggestions for news headlines and answer financial questions.
Reuters editors in Latin America are using it to tidy up copy for journalists writing in a second language.
Newsquest is using a chatbot to send Freedom of Information requests.
KSAT-TV uses AI to ingest an excerpt of video and summarise it as text in its CMS.
WFMZ-TV uses GenAI to monitor incoming e-mails and create events it places on the editors’ planning calendar after classifying them as Worthy, Unworthy, or Unsure.
Núcleo uses it to monitor and interpret government documents.
But we have also seen bolder use cases where publishers have put GenAI directly in front of their consumers. For example:
Kölner Stadt-Anzeiger Medien created a virtual journalist called Klara Indernach that writes more than 5% of its articles.
India Today has an AI news anchor.
El Vocero de Puerto Rico uses AI to quickly translate and immediately publish alerts from the National Hurricane Center.
The BBC uses it to present content in different formats and test them with different audience segments, as well as to create explainers based on older content.
Reuters has built its own large language model for clients of its legal service to generate answers about cases, statutes, and regulations.
What are we going to see in the year ahead? Here are some themes that I expect to emerge:
Quick wins vs. long-term impact. Most news organisations have simply grabbed the low-hanging fruit so far. Since we all have limited resources, this is the year when we start to debate whether it is worth sacrificing long-term gains for short-term wins.
Questions over ROI. Many news organisations are jumping on the bandwagon to see what they can use GenAI tools for. This is the year when we move beyond FOMO and start asking exactly how much money they are saving us or making us, and at what cost.
The build vs. buy debate. There are now GenAI tools out there that can help news organisations move quickly. Is it worth building your own? How about overcoming the Not-Built-Here culture that still exists in many companies?
Licensing deals. Axel Springer and the Associated Press have signed licensing deals with OpenAI, and News Corp is apparently in talks as well. Apple is reportedly having similar discussions with news producers as it develops its own GenAI tools.
Lawsuits over copyright. You’ve seen The New York Times’ lawsuit that questions whether GenAI companies can use their content as training material without compensation. The outcome will help determine how many more we see.
Business models in flux. Search engines will soon be presenting bullet points summarising the news. How many readers will click through to our sites? What will that mean for business models predicated on advertising impressions, affiliate revenue, and subscription revenue?
Regulations. News organisations will be in a hurry to implement GenAI, but their legal departments will slow the process down because they do not have the security and data privacy framework in place to move these initiatives forward quickly. This may be further slowed down by nascent government regulations.
New audiences for old newspapers. Multilingual news from unilingual reporters; video and audio from traditionally text-based news organisations: GenAI makes it cheap and easy to get into busy consumers’ ears and younger consumers’ phones.The next question will be how to monetise them effectively.
Robust data strategy. Almost all chief data officers believe that getting a strong data foundation in place is vital for using GenAI — but only 38% have what they need in place. Garbage in, garbage out.
If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.