News automation is becoming more common in UK journalism, sparking discussions about how the use of artificial intelligence (AI) technologies in the newsroom might change the relationship between journalists and the public (Smith 2024). This trend is also reflected internationally, as, driven by economic pressures and the general AI hype, news organisations across the world have been exploring how automation may support the productivity of journalists (Beckett and Yaseen 2023). However, there is a concern that the uncritical use of these technologies could decrease public trust in journalistic institutions (Newman et al. 2024).

AI is often used as an umbrella term that includes rule-based automation and more advanced machine learning-based systems, essentially describing the ‘automation of tasks or decisions (either fully or partly) that would previously have required the intelligence of a human being’ (Schjøtt Hansen et al. 2023, 17–18). While rule-based systems have been employed in news production for a while (Thurman 2019), recent advancements in machine learning have accelerated AI use in news production workflows (Simon 2023; Esposito 2022).

In the UK, both private and publicly owned media have incorporated AI into their newsroom workflows. This includes the automation of news production at organisations such as Reach PLC and the BBC (Stalph et al. 2023), to, for instance, generate local data-driven stories, as well as personalise news distribution (BBC 2024b). Collaborations between technology companies and UK news organisations, like the partnership between AI developer OpenAI and the Financial Times, have emerged to develop AI solutions tailored to specific journalistic needs (Reuters 2024). The growing prevalence of AI in UK journalism has also prompted the creation of editorial guidelines for the professional use of AI tools, exemplified by initiatives at the Guardian (de Lima-Santos et al. 2024) and the BBC (BBC 2024c).

These developments suggest that news automation might be becoming an integral part of journalistic workflows in the UK. Consequently, this survey sought to investigate whether UK journalists were aware of the use of some types of news automation in their editorial environment and how this awareness may affect other professional experiences that have been said to be associated with the use of automation, such as increased editorial freedom and decreased job security (Lindén 2017; Flew et al. 2012). Therefore, in this 2023 survey, the UK, alongside other countries participating in the third wave of the Worlds of Journalism Study, introduced two new questions. These questions pertained to journalists’ awareness of (1) the use of automation for news text production¹ and (2) the application of personalised news distribution² within their newsrooms. At the time of the survey’s composition (2019), these were among the most prevalent editorial uses of automation in news organisations.

4.1 The prevalence of news automation in UK newsrooms and its drivers

The findings show that only 7% of UK journalists were aware of working in a newsroom that used automated text production, and only 10% of working in one where personalised news distribution was used (see Figure 4.1). These figures suggest that a minority of UK newsrooms have incorporated such technologies into these particular workflow stages. However, it is important to note that these data solely reflect the journalists’ awareness of the use of these technologies. The use of automation in these editorial processes that happens unbeknown to the surveyed journalists remains unaccounted for in the data.

Figure 4.1

The propensity to automate these two editorial processes varies by the media organisation’s ownership. Specifically, 9% of journalists primarily employed by private media organisations worked in newsrooms where automated news text production was used. By contrast, only 3% of journalists primarily working for outlets with public ownership reported that the technology was used for text production in their newsrooms (see Figure 4.1).

Private media organisations face substantial pressure to innovate and remain competitive in an ever-evolving digital news landscape. This dynamic may explain the somewhat higher prevalence of automated news writing in these newsrooms compared with publicly owned ones, which enjoy a larger degree of independence from market-driven competition and may, therefore, not be as reliant on the efficiency gains in news production that AI enables. In the ongoing pursuit of faster, more efficient, and cost-effective news production, automated solutions may offer an enticing opportunity – once integrated into editorial processes – to increase the speed and volume of publishing while saving resources (Diakopoulos 2019). News organisations that rely on their ability to publish news quickly, such as news agencies, may particularly benefit from the efficiencies afforded by automated news text production (Thäsler-Kordonouri and Barling 2023). Despite other studies indicating that ‘both public and private funding models are under pressure as audiences shift their attention further towards digital channels’ (Newman et al. 2024, 64), the 2023 survey’s findings suggest that privately owned media organisations exhibit a slightly but significantly higher propensity to alleviate this pressure through the economic advantages promised by automated text production than publicly owned ones.

However, research indicates that readers find news stories produced with the help of automation harder to understand than those fully authored by humans (Thäsler-Kordonouri et al. 2024). Furthermore, the automation of news production may pose risks to the maintenance of journalistic values, given that AI-assisted workflows often lack the transparency inherent in human processes (Cools and Koliska 2024; Diakopoulos and Koliska 2017). Indeed, an international study revealed that many journalists remain ‘concerned about the ethical implications of AI integration for editorial quality and other aspects of journalism’ (Beckett and Yaseen 2023, 7). These and other considerations may explain why publicly owned media exhibit a lower propensity to automate their workflows in this way. Some public service media have stringent AI implementation standards in news text production. This is exemplified by the BBC’s ‘AI Principles’, which state that the ‘use of AI will reflect the public service mission and values of our organisation. When we use AI to create, present or distribute content we will make sure that this complies with the BBC editorial values, guidance and guidelines’ (BBC 2024c).

The survey shows a different relationship between the use of personalised news distribution and the ownership of media organisations. Specifically, only 8% of journalists mainly employed by private media organisations worked in newsrooms where personalised news distribution was used, whereas 22% of journalists in publicly owned outlets reported the use of such technology in their newsrooms (see Figure 4.1). Automating this workflow step supports the targeted delivery of relevant news content to specific audience groups. This functionality may help publicly owned news outlets, such as the BBC, to fulfil their mandate to serve diverse audiences across the UK by highlighting relevant content to various demographic segments (BBC 2024c).

Another organisational characteristic pertinent to the implementation of these two types of news automation is the news outlet’s media cultural background: whether it has, for instance, a newspaper, radio, or television tradition (see Chapter 3). The findings of the 2023 survey indicate that the outlet’s media cultural background significantly affects newsrooms’ adoption of automated text production.

The proportion of journalists who indicated that automated text production was used in a newsroom where they worked was highest among those who worked mainly for news agencies (33%) (see Figure 4.2). News agencies, in particular, may benefit from the resource savings and increased speed that automated news writing can provide, as rapid publishing is essential to their business model. In the UK, the news agency RADAR (Reporters and Data and Robots) uses this production approach as its business model, describing itself as ‘the world’s only automated local news agency [that combines] human analysis and writing skills with cutting-edge automation tools’ (RADAR 2024).

Figure 4.2

4.2 The promises and perils of news automation

The introduction of automation into the news cycle has sparked debate regarding the benefits and risks associated with outsourcing editorial processes to these technological systems. One anticipated benefit of AI is the potential to relieve journalists of routine tasks that can be easily automated, thereby allowing them to focus on more substantive and meaningful responsibilities, such as covering more complex topics in their news reporting (Beckett and Yaseen 2023; Flew et al. 2012).

However, the 2023 survey found that the editorial freedom³ felt by journalists who worked in newsrooms where, to their knowledge, either of the two types of news automation were used was slightly but significantly less than for journalists who worked in newsrooms that did not use these technologies.

Specifically, among journalists who reported working in newsrooms where automated news text production was not employed, 67% felt they had ‘complete’ or ‘a great deal of’ freedom to select which stories to cover. In contrast, this feeling was shared by only 55% of journalists who reported working in newsrooms where automated news text production was used. In newsrooms that used personalised news distribution, 58% of journalists felt they had ‘complete’ or ‘a great deal of’ freedom to select which stories to cover compared with almost 69% in newsrooms that did not use this technology (see Figure 4.3).

Figure 4.3

Figure 4.4

Considering a risk associated with AI implementation, job security,⁴ the survey shows a significant relationship between awareness of the use of these two types of automation in the newsroom and journalists’ fear of job loss. Previous research has linked the use of AI for editorial tasks with insecurity among journalists regarding their professional roles, with concerns about being replaced by the technology emerging as a recurrent theme (e.g. Lindén 2017; Graefe 2016). The 2023 survey’s findings add empirical weight to these concerns.

Among journalists working in newsrooms where news writing was automated, 37% expressed concern about losing their jobs in journalism within the next 12 months.⁵ In contrast, this concern was shared by only 27% of journalists in newsrooms where this was not the case. This disparity is even more pronounced with respect to the use of personalised news distribution: in newsrooms that used this type of automation, 48% of journalists indicated being worried about job loss within the next year, compared with just 26% in newsrooms that did not automate this task (see Figure 4.4).

The fear of job loss and perceptions of editorial freedom are undoubtedly influenced by a range of factors beyond AI usage in the newsroom. However, further preliminary analysis shows that, even when controlling for other potentially relevant organisational variables such as the outlet’s media cultural background, its reach (local, national, or international), and primary ownership, both the use of automated news writing and personalised news distribution in newsrooms where journalists work, contribute significantly to journalists’ fear of job loss and perceptions of editorial freedom. However, it is important to be clear that with cross-sectional data such as this, it is not possible to confidently identify causal relationships.

4.3 Conclusions

This chapter has presented and analysed UK journalists’ awareness of the use of two forms of automation within newsrooms – automated text production and personalised news distribution – and explored the relationship between use and organisational characteristics – ownership and media cultural background – and journalists’ perceptions of editorial freedom and job security.

The findings suggest that these AI technologies have yet to achieve widespread adoption in UK newsrooms. However, certain integration tendencies already exist: private media organisations and those with a news agency background are more likely to use automated text production. Publicly owned media organisations are more likely to use personalised news distribution. These tendencies can be partly attributed to the business models and publication logics of these media organisation types.

Despite the limited uptake of these forms of AI in UK newsrooms, this survey’s results indicate that their implementation in the newsroom significantly contributes to journalists’ perceptions of editorial freedom and job insecurity. These relationships have been only briefly touched upon in this chapter, but the preliminary findings underscore the necessity for more comprehensive research, particularly in the context of the increasing integration of other forms of AI into newsroom workflows in the UK and globally.

Footnotes