Journalism around the world has been under attack for decades, with public trust in the media steadily declining—an issue that has only worsened in recent years. While some factors are outside the control of journalists, others are of our own making.

Many media organisations worldwide have fallen victim to sloppy reporting, publishing unverified claims, and putting out subpar content, often due to shrinking revenues, understaffing, and numerous other challenges. Yet, despite the pressure, journalism cannot afford to give up.

For journalism to regain the trust of the public, new solutions must be found. One such potential tool is Artificial Intelligence (AI).

While the suggestion may initially seem unorthodox, I believe AI, when used correctly, has the potential to help make journalism more balanced, less biased, and ultimately more trustworthy.

Artificial Intelligence is still a relatively new concept for the general public, often associated with uncertainty, fear, and scepticism. But by automating mundane tasks and providing valuable insights into audience preferences, AI can help reporters and editors focus more on producing well-researched, impartial, and contextualised content.

At the University of Malta, researchers in the Department of Artificial Intelligence under the supervision of Dr Dylan Seychell are exploring how AI can be leveraged to tackle bias in journalism.

One such project, NBxAI (Exploring Visual Bias in News Content Using Explainable AI), is particularly focused on helping journalists identify and address bias in news coverage. 

Having been involved as an advisor in this project, I’ve seen firsthand how AI can offer valuable assistance to newsrooms.

Tackling Bias with AI

Gabriel Hili is a researcher building these tools to help journalists detect and reduce bias in their reporting.

When I sat down with Hili to discuss the project, I asked him what "bias" meant in the context of journalism. He explained that there are many forms of bias, both intentional and unintentional. Bias, he noted, is not always a result of malicious intent.

In fact, many of the biases we encounter in media come from a lack of time or resources, as well as the overwhelming demands placed on journalists.

"Most Maltese newsrooms have become too small, and the workload too large for journalists to effectively monitor their own content," Hili explained. "As a result, biases—intentional or otherwise—often slip through the cracks."

Hili and the team gathered insights on the realities of newsroom life by speaking with several current and former journalists, including myself. These conversations helped inform the development of tools to combat bias in journalism.

News Align: A tool for detecting bias

One of the most exciting aspects of the NBxAI project is a tool called News Align, which aims to be an invaluable resource for journalists looking to detect and correct bias in their reporting.

News Align is a bias-sensitive news aggregator. For those unfamiliar with the term, a news aggregator is a tool that collects and organises articles from various news sources, such as Google News. Most aggregators, however, fail to account for potential biases in the way news is presented. While some aggregators highlight whether a media outlet leans left or right politically, these services are generally limited to international news.

As Hili explained, News Align fills a gap in the Maltese media landscape.

"News Align is designed specifically for Maltese news," he said. "It aggregates articles on a particular event, allowing users to see how different Maltese websites report the same story. It compares not just the text, but also the headlines, images, and how relevant those images are to the story."

But News Align isn’t just about highlighting bias in text. It also addresses visual bias - a subtle but powerful way that media can shape public opinion. For example, some media outlets may use unflattering images of political figures or marginalised groups, such as migrants, influencing the viewer’s emotional response.

It’s not only about intentional bias. A journalist might, out of expediency or a lack of stock photos, publish an article with a photo that does not really reflect the headline. This system will help journalists identify these shortcomings and address them.

News Align also includes an AI-powered chatbot that provides further context on a news event. The chatbot pulls information from various local news articles on the subject, helping to counteract bias from any single source. 

While News Align is still in development, the team hopes to make it available to both journalists and the public in the near future once all the ethical and safety considerations are tested

 A growing dataset for accuracy

News Align relies on a constantly updated restricted database of Maltese news articles, which is refreshed every 15 minutes. This ongoing data analysis helps ensure that the tool remains accurate and up to date.

An example of a database query finding links between reports.

The system also learns over time, becoming more accurate with each use. However, machine learning needs a starting point, and in the case of News Align, that starting point was human input. A total of 144 participants were asked to rate how relevant images were to the headlines of 100 articles. Their feedback helped train the AI model, ensuring it reflects how Maltese people interpret news visuals.

While the system is already about 64% accurate, there are plans to improve its performance by adding more context-specific capabilities, such as face recognition, which would allow the system to identify key figures in photos (e.g., Prime Minister Robert Abela or Opposition Leader Bernard Grech). This will help refine the tool’s ability to assess bias, making it even more useful for journalists and media consumers.

Other AI projects in the works

The team is also working on a project that involves AI technology in news broadcasts. This system can track how often a particular person appears on screen and for how long. This could be useful in situations where there are accusations that one individual is receiving disproportionate coverage. In the context of political campaigns or controversial figures, such a tool could help verify claims about biased airtime.

Gabriel Hili.

Another project being developed by the team in the Department of AI involves tracking eye movements to determine which parts of an article or webpage attract the most attention. By understanding where readers focus their gaze, journalists can optimise their content for better engagement.

Other researchers and students under Seychell’s supervision are working on other interesting projects, including research into the use of emojis in news articles and how they impact the emotional tone of reporting.

The NBxAI research team showcased the project to media professionals

Collaboration with industry stakeholders

A key element of this research initiative is its collaboration with journalists and media professionals. In May 2024, the research team in collaboration with podcaster Jon Mallia showcased these projects to a group of notable media figures, including Times of Malta Editor-in-Chief Herman Grech, assistant editor Mark Laurence Zammit and Lovin Malta’s David Grech Urpani. The feedback from these industry professionals was overwhelmingly positive, and the event sparked an engaging conversation about the future role of AI in journalism.

Hili emphasised the importance of collaborating with journalists and media outlets. "Getting feedback directly from industry players is invaluable," he said. "It helps us understand the practical challenges journalists face and ensures that the tools we develop are actually useful in real newsroom settings."

Neil Camilleri.

The Department of AI is also collaborating with the Faculty of Media and Knowledge Sciences. The annotation data from the study has been shared with Professor Vince Briffa, who supports the NBxAI team in compiling observations that reflect the insights that came out of this project with the aim of avoiding bias.

AI holds great promise for the future of journalism. By helping journalists identify and address bias - both intentional and unintentional - AI tools like News Align can contribute to fairer, more balanced reporting. As the NBxAI project continues to evolve, it has the potential to transform how journalists approach their craft, while also empowering media consumers with the tools they need to make more informed decisions about the news they consume.  The research team is publishing updates on the website.

Ultimately, AI can be a valuable ally in journalism’s ongoing struggle to regain public trust. If used responsibly and collaboratively, it can help media outlets overcome many of the challenges they face today—allowing journalists to focus on their core mission: delivering the truth, impartially and fairly.

 The project ’Exploring Visual Bias in News Content using Explainable AI’ (NBxAI) is financed by the Xjenza Malta, for and on behalf of the Foundation for Science and Technology, through the FUSION: R&I Research Excellence Programme.

Neil Camilleri is a freelance journalist and former Editor-in-Chief of The Malta Independent. He is also an advisor on the AIxNB project.