Ensuring AI is enhancing-not undermining- journalism

We live in a rapidly evolving digital landscape, where terms such as Artificial Intelligence (AI) or Large Language Models (LLMs) have been increasingly heard and can potential revolutionize various industries. What happens with journalism?

Well, we have been experiencing the advent of AI in journalism for quite some time, not knowing or grasping how fast it will de developed in what we know today.

Automated news writing and distribution, content recommendation algorithms, and data journalism have been around massively in the last few years and are terrific examples of AI's growth.

However, it wasn't until the advent of LLMs like GPT-3 that the trend became fully apparent and opened new possibilities, challenges and discussions.

What are LLMs?

They are essentially trained algorithmic models that can generate human-like text, which is the raw material for any application in journalism.

The more prominent example until recently was the generation of automated reports, particularly of those that were based on structured data such as financial reports or sports scores.

It was the first indication of how efficiency in a newsroom can be increased, by allowing journalists focus on less redundant and more meaningful investigative tasks.

However, it turns out LLMs can also adapt news content to what individuals want to read, based on their preferences and history, which in turn can strengthen engagement and loyalty.

They can also assist fact checking functionalities by offering cross-referencing information to fact checkers from various sources; a key step in the battle against disinformation, which is essential in maintaining the integrity of the profession.

Interactive journalism can also be benefited by LLMs, with chatbots providing updates or answer specific readers' questions about a topic. Forbes recent launch of its generative AI search platform built with Google Cloud is a great example of how an established newsroom can provide personalized searches for its readers.

The potential of LLMs in journalism is obvious and exciting, but on the flipside it also raises certain ethical considerations that need to be taken into account.

These models can create grammatically correct and coherent text but this doesn't mean that they inherently understand the content they're generating, which can lead to inaccuracies; a particularly problematic aspect of any journalistic process.

Objectivity of the news content can also be undermined by the fact that like any other AI based system, these models reflect biases that are engraved in the training set they were built on.

So, as we move further into this AI-driven future of journalism, it's imperative to be able to balance between the need to leverage the potential of LLMs and to address these key challenges, that require academics, journalists, technologists and policymakers to collaborate effectively and in a timely manner.

But in this collaborative scheme, journalists should not take a back seat. We are responsible for guiding the development and deployment of LLMs in journalism in a way that accuracy, fairness, and transparency are protected.