Expert David Caswell on why generative AI may transform the news ecosystem and how journalists and news companies should adapt

Innovation in journalism is back. Following a peak of activity in the mid 2010s, the idea of fundamentally reinventing how news might be produced and consumed had gradually become less fashionable, giving way to incrementalism, shallow rhetoric and in some cases even unapologetic ‘innovation exhaustion.’ No longer. The public release of ChatGPT in late November of 2022 demonstrated capabilities with such obvious and profound potential impact for journalism that AI-driven innovation is now the urgent focus of the senior leadership teams in almost every newsroom. The entire news industry is asking itself ‘what’s next’?

For many people in journalism the first half of 2023 was a time for asking questions and learning the basics of AI. What can ChatGPT actually do? What is generative AI? What is a language model? What is a ‘prompt’? How dependable are these tools? What kind of skills are required to use them? How fast is this technology improving? What are the risks? How much of all this is just hype?

Many newsrooms went further, providing their employees and audiences with statements or guidelines describing how they intended to approach the use of generative AI in their workflows and news products. Some even began publishing a few experimental articles written by ChatGPT. Very few, however, have yet taken specific steps to pragmatically and routinely apply these technologies in their newsrooms. Change is in the air, but specific initiatives are harder to find.

Over the past six months I have had the privilege of spending significant time discussing AI with the senior leadership of more than 40 news organisations, ranging from scrappy digitally-native newsrooms in Asia, the Middle East, and Latin America to many of the largest global news providers based in the US, UK, and Europe.

This access came as a result of my familiarity with AI-driven innovation in news, acquired from a combination of applied product innovation at large media companies in the US and UK, and from a small parallel academic career exploring the first principles of computation applied to news production and consumption.

My journey with many of the questions now facing news organisations began in Silicon Valley more than a decade ago, and my exploration of the GPT family of language models began in 2019 at the BBC. My career focus — a relative niche until the arrival of ChatGPT — was suddenly of great interest to many people in the news industry.

This article is based on a distillation of what I have learned from these conversations. It assumes a knowledge of generative AI’s general capabilities and potential, and examines some of the ways that large news organisations are thinking about its strategic and practical implications for their newsrooms. My intent here is to help advance the conversation beyond awareness and towards specific initiatives that can help move newsrooms forward in preparing for an uncertain AI-mediated future.

Any point of view regarding the application of AI to news is somewhat speculative, especially given the remarkable pace of advancement in AI functionality. Nonetheless, I believe that some clarity may be emerging.

Coming up with a list of things that your newsroom might be able to do using ChatGPT is fairly easy (Summarisation! Rewriting in a simpler style!). It is much harder to clearly identify exactly what it is that you are trying to achieve with generative AI, why you are trying to achieve it, how you might plausibly achieve it at scale in routine and professionally managed operations, and whether that achievement will even continue to be relevant as AI fundamentally alters the competitive landscape in the coming months and years.

To frame this discussion I will begin with an overview of potential strategies for using AI in news, before turning to options for the practical deployment of AI in newsrooms, and the infrastructural and organisational requirements needed to support those options. At the end of the piece I will offer a point of view for changes that we might anticipate for the news industry in the longer term.

1. Applying generative AI to news: product and editorial strategies

Efficiency-focused strategies

The most obvious opportunities for applying generative AI to news are in bringing new efficiencies to specific and familiar steps within the existing news production workflows supporting an organisation’s existing news products. This ‘more-efficient-production-of-existing-products’ strategy is attractive in its simplicity, but its benefits will almost certainly be short-lived because it assumes that the existing media environment will continue in roughly its existing form.

It is increasingly possible that the competitive environment, product offerings, production workflows and business models of news organisations will change, possibly radically, as use of generative AI becomes ubiquitous and as AI-based media products appear.

Continue reading

by David Caswell

Photo by Cash Macanaya on Unsplash