10 Ways Journalists Use AI Tools in the Newsroom

Introduction

Many industries use Artificial Intelligence to increase productivity; news organisations are no exception. Companies are creating digital figures like JAMES (Twipe) and Klara Indernach (Express.de) to mix the computational abilities of AI with human creativity. These tools make tasks easier and offer chances for personalised reader engagement, rigorous fact-checking, and interactive content creation. Let’s have a look at how AI is currently used in newsrooms.

1. News article generator

Generative AI can be used in newsrooms to create high-quality texts and reports, reducing the time spent on it by journalists.

EXPRESS.de has recently introduced Klara Indernach (KI), an advanced artificial intelligence system, to their newsroom. Klara is a sophisticated tool that can structure texts, conduct extensive content research and summarise vast amounts of information quickly and efficiently. These capabilities greatly enhance the speed and scope of content creation at EXPRESS.de, especially when it comes to articles with predictable content to write, such as sport reports.

Human editors still play an active role in the publication process, and they must review every piece of content that Klara contributes, ensuring the integrity and authenticity of the news. This collaboration guarantees that the integrity of journalism is still upheld, while also benefitting from the productivity gains that AI can provide. Although Klara has taken on repetitive tasks, EXPRESS.de remains committed to the values of traditional journalism. The German media industry has voiced strong opinions on the matter; however, there is ongoing debate about whether readers care about AI-generated content. Discussions from #DGS2023 suggest that most people do not notice whether AI was used to create content. While many companies using AI prefer using disclaimers on AI-generated articles, Arist von Harpe offers an interesting perspective:

We do not highlight AI-aided articles. We’re only using [AI] as a tool. As with any tool, it’s always the person using it who is responsible for what comes out.

Arist von Harpe

2. Knowledge base with AI

Large corporations such as Google, Microsoft and OpenAI are interested in the archives of media organisations to use such data to train their Large Language Models. While allowing access to their archives may be of benefit for media organisations, many are having issues with allowing free access to their archives.

Mediahuis is exploring potential solutions to this by collaborating with ML6 to utilise Retrieval Augmented Generation (RAG) techniques to create an internal knowledge base at a fraction of the cost that training a full LLM would require. This technique may be particularly beneficial for Mediahuis to maintain the integrity and accuracy of its content while driving innovation in AI-driven journalism.

Mediahuis is in dialogue with tech giants like Microsoft, ML6, and Google to explore models where their data can be used to train LLMs, leading to the development of more domain-specific language models. Such collaborations can pave the way for a new era of AI-driven journalism, where compromises between media organizations and tech companies leads to the continual improvement of AI technologies in this domain.

3. Generate news quizzes

The 2023 Reuters Digital News Report highlights the appeal of games, including news quizzes, with 25% of subscribers to newspapers in the U.S. citing them as a reason to purchase a subscription. Several news organisations are trying to pick up on this trend. Several news organisations have seen potential in such a solution and have started developing automatically generated news quizzes based on existing articles. TIME ran a series of experiments, utilising ChatGPT to mine its archive of 200 million words of iconic stories to gauge current affairs knowledge for quiz content. This AI integration combines engagement and education, encouraging readers to pay more attention while reading news.

However, this approach cannot be completely automated as AI is still subject to a phenomenon known as “hallucination”, where AI mixes accurate and misleading information in a way that is difficult to detect. To achieve accurate results, precise prompting and having a human oversee the process is critical.

4. AI image generators

Workers in the media industry can use artificial intelligence to create images, illustrations, and infographics for news stories. These tools help the design process, which usually requires time and expertise, and can produce relevant visuals by inputting text or data. This speeds up content production and saves human resources. It’s also a cost-effective solution, especially for smaller news outlets that cannot afford a large design team or expensive freelancers. 

AI-generated visuals can be quickly produced and personalised in marketing, enhancing advertisement campaign relevance and audience engagement. The images generated by AI offer opportunities for creative storytelling, with real-time visuals enhancing narratives and audience engagement.

5. AI voice generators and AI transcripts

One of the cornerstone applications of AI in a newsroom is the transcription of audio interviews or discussions into written content. This transformation is crucial as it facilitates the archival and easier dissemination of information. For instance, journalists often engage in verbal discussions or interviews, which are rich in content. Leveraging AI-driven transcription services, these verbal exchanges can be swiftly converted into text. This makes the content readily available for editing, publishing, or sharing across different platforms, enhancing the newsroom’s ability to repurpose or reference past content.

With the advent of AI, the efficiency and accuracy of transcription processes have seen a tremendous improvement, with tools like VG’s Jojo. Unlike manual transcription, which is time-consuming and prone to errors, AI-enabled transcription services present a faster and more accurate alternative. Utilizing sophisticated algorithms, they can accurately decipher speech, even in noisy environments or with multiple speakers. This significantly reduces the time from recording to publishing, accelerating the news production cycle. It also frees up journalists and editors to focus on more creative and analytical aspects of news production.

Building on this, AI-powered text-to-speech technology has made impressive advancements in speech synthesis, enabling realistic voiceovers in a minimal timeframe. This technology also allows for accurate translations that preserve the original intonation, thereby enhancing the authenticity of news reporting. These advancements pave the way for innovative applications like real-time translations and automated dubbing. Such innovations make content more accessible and engaging to a global audience, thus broadening the reach of newsrooms.

By integrating AI-powered audio transcription, speech-to-text, and text-to-speech technologies, newsrooms can significantly enhance their efficiency, accuracy, and content quality. This amalgamation streamlines workflow, helping newsrooms meet the demands of digital media. Additionally, AI aids in the creation and distribution of multimedia content, leading to a more informed and engaged audience. Through these technological advancements, newsrooms are better equipped to deliver high-quality content across various platforms to wider audiences.

6. Personalised news: JAMES

One of the most interesting uses of AI in the newsroom is the possibility to offer individually tailored content to readers and subscribers. A great example of content personalisation is JAMES, a digital butler created by Twipe in collaboration with The Times to help news publishers enhance reader engagement through personalised emails to serve the evolving needs of modern news consumers. JAMES personalises the distribution of newsletters by learning from reader behaviours and preferences, tailoring the content to individual reader habits​​. 

Personalised newsletters are becoming increasingly popular in the news industry as they use data and machine learning to improve reader engagement and retention. With tools like JAMES and in-house personalisation efforts, news organisations are better equipped to cater to the diverse interests of their readers while also achieving their digital growth targets.

7. AI Article Summaries

AI can be used to generate titles, meta descriptions and summaries of news articles.

1. Quick Summaries: AI can generate summaries that provide readers with an overview of the main points in an article. This feature is similar to the one introduced by the news app Artifact (see GIF).
2. Customised Styles: AI can create summaries that are tailored to different audiences, using unique styles such as “explain like I’m five” or emojis to make them more engaging.
3. Time-Saving: By summarizing lengthy articles, AI can save users time, which is especially useful in the fast-paced news industry.
4. Accessibility: Making news more accessible is crucial, and AI can play a role in achieving this by providing summaries that are easy to read and understand. This can broaden the audience base and make information more inclusive.

8. AI Fact-checking

Fact-checking is a time-consuming and expensive task for humans to perform. With the proliferation of online misinformation and fake news, readers increasingly value accuracy in their news sources. AI algorithms can rapidly cross-check data and validate information from different sources, making spotting potential inaccuracies or lies easier.

These systems can compare claims against vast databases of trustworthy information, making the fact-checking process quicker and more efficient than if done solely by a human.

However, achieving optimal accuracy in fact-checking requires a balanced approach that combines automated AI processes and human judgment. While AI can accelerate initial assessments and flag potential issues, it may not fully understand the context or implications of a statement. Human fact-checkers are still necessary to provide critical thinking skills, topic expertise, and a deeper understanding of context to ensure accuracy.

By working together, AI and human fact-checkers can deliver trustworthy and precise fact-checking results. This collaboration allows for the speed of technology to be combined with the interpretative capabilities of humans, resulting in an effective strategy for fact-checking.

9. Comment moderation

AI can assist with monitoring online discussions and identifying inappropriate content across various media platforms. With ever-increasing internet activity, it has become challenging for human moderators to keep pace with the sheer volume of user- and automatically-generated content.

AI algorithms, particularly those using Natural Language Processing, excel at scanning text and identifying patterns that may indicate hate speech, harassment, misinformation, or other inappropriate content. When such content is detected, AI systems can flag it for human review or, in some cases, automatically remove it to maintain a safe and respectful online environment.

Ensuring open yet respectful dialogue on media platforms is a complex challenge, as it requires a delicate balance between freedom of expression and the need to prevent harm or toxic behaviour. AI can assist in achieving this balance by serving as a first line of defence against inappropriate content.

One pitfall to note is that AI assessments can be inaccurate, sometimes leading to both false positives and negatives. Therefore, human moderation remains crucial for more nuanced assessments, contextual understanding, and addressing evolving forms of inappropriate content. Media organisations must once again find a delicate balance between investing in human moderators and relying on AI-based moderation tools.

10. Chatbots

As referred above about Mediahuis’ collaborative efforts with Microsoft, ML6, and Google, data is being used to train LLMs in pursuit of pioneering chatbot technology.

The chatbot technology, while offering efficiency and convenience, faces notable challenges. First, ensuring accurate natural language understanding remains a significant hurdle. Chatbots must comprehend diverse user queries, accounting for nuances, idioms, and context.

Contextual understanding is equally vital to maintain coherent conversations, with chatbots needing to remember and reference prior messages to provide relevant responses. Personalisation presents another challenge, balancing the provision of personalised responses based on user data with privacy concerns and data protection regulations.

Handling unexpected and ambiguous queries further tests chatbots’ adaptability. It must respond to unexpected user inputs, asking clarifying questions or indicating limitations when necessary. Additionally, it has to maintain support for multiple languages, which are complex and require accurate translations and context-aware responses.

Ensuring secure integration with backend systems is important, preventing data leakage or security vulnerabilities. Furthermore, ethics and bias are paramount; chatbots must be monitored to avoid biases and offensive responses. Ongoing maintenance, user trust-building, and striking the right balance between automation and human intervention are essential for the successful deployment of AI chatbots.

Conclusion

AI tools have changed journalism by combining storytelling with data analytics. This collaboration makes journalism more efficient and creates a better experience for the audience. Even though AI can undertake many mundane tasks, investigations and storytelling requiring human creativity remain at the heart of journalism. AI has expanded the possibilities for journalism, just like photography did for painters. While AI blurs the line between journalist and tool, storytelling remains the focus. The future of journalism is a duet between journalists and AI, with the enduring spirit of inquiry and storytelling guiding the way.