There is both a perceived and material gap in AI adoption between large national news organisations and smaller local news providers (Rinehart & Kung, 2022). This gap has empirical indicators and has been examined in the scholarly literature, with some estimates suggesting that only 10% of AI systems deployed in journalism are used in local newsrooms (Aubin Le Quéré & Jakesch, 2022). Several factors help explain this imbalance: large national and international publishers typically have the financial capacity to develop in-house AI models, while smaller regional and local outlets are more likely to rely on off-the-shelf tools provided by platform companies (Simon, F. M. 2024b). Current debates about AI in journalism tend to centre on larger organisations, leaving local news organisations with less sustained attention despite their heightened exposure to these changes (Simon, F. M. 2024a).

The Rojas case is invoked here not for its emotional weight but as a single, verifiable event that transformed a contested methodology into an authoritative institutional standard. On 29 June 1892, the people of Necochea, a small town in Buenos Aires, woke to horror. Two murdered children, but their mother, Francisca Rojas, was found injured and alive. Police rushed to the scene and arrested a neighbor based on Rojas’s accusations. At first, the case seemed straightforward, with a suspect and a witness.

Imagine a mental health support app powered by artificial intelligence (AI). Its developers notice the app recommends professional help more aggressively to women than to men, a clear gender bias. They fix it, and they succeed. But a few months later, something else surfaces: older users are now receiving shorter, more dismissive responses than younger users for the exact same struggles. Nobody changed anything related to age. So what happened? What happened is the bias spillover effect, scientifically defined as “the unintended alteration of behavior on one social axis when mitigating another” (Mijalli et al., 2023).

OpenAI recently released a new product designed to support medical care: ChatGPT Health. They describe this innovation as an “experience that securely brings your health information and ChatGPT’s intelligence together, to help you feel more informed, prepared and confident navigating your health” (Introducing ChatGPT Health, 2026). They claim that health is already one of the most common ways people use ChatGPT, and that this platform will increase users’ knowledge, preparation and confidence in managing their health (Introducing ChatGPT Health, 2026). However, while AI-powered health tools may reduce information overload and improve accessibility, their integration into help-seeking processes raises significant ethical questions regarding misinformation, autonomy and bias.

For decades, human-computer interaction was built on a comfortable assumption: people click, systems respond and the boundaries are clear. A button is a button, a menu is a menu. When things go wrong, you can often spot the missing step, the frozen cursor, the bad modal window.

For years, the industry has focused on making models bigger. This training time scaling (Kaplan et al., 2020) made models highly fluent, similar to a student who memorised the entire textbook. But fluency is not the same as reasoning. Large language models (LLMs) still struggle with complex logic, maths or coding tasks because they respond too quickly, predicting the next word without truly thinking (McCoy et al., 2023).