AI good for journalism

The last two decades have put pressure on journalists, editors, and newsrooms [1], [2]. On the content side, young digital natives have different news habits from the older media consumers . They rely more on alternative and free information sources and are less likely to pay for news subscriptions , [3]. Other segments of the population shun mainstream media due to perceived political bias and distrust in authorities . These attitudes may become exacerbated by co-ordinated disinformation campaigns [4] and amplified in sealed information enclaves, such as online echo chambers [5] and alleged search and recommendation bubbles [6], [7]. Some population segments also lack the language, reading and other skills that are required to appreciate quality news. They may experience news fatigue or exhibit news avoidance [8], [9]. On the business side, media income from advertising, subscriptions and sales has dropped due to the availability of free online news sources, social media, search engines, and other intermediaries , . In particular, social networking sites exploit their platform power by leveraging information about consumer behaviours to offer individually targeted promotion at lower prices than traditional mass media [10]. As broken business models lead to newsroom layoffs , these challenges become even harder to tackle, creating a vicious cycle. For example, in the US, more than a quarter of all newsroom jobs were lost between 2008 and 2020 [11]. Similar developments have been experienced elsewhere .

At the same time, many of the same mechanisms that put journalists, editors, and newsrooms under pressure make quality journalism more important than ever. Rigorous evidence-based journalism that is based on robust democratic values and open about its positions and values can help to defuse growing populism and distrust in authorities. Indeed, many analyses1 show that countries, such as Norway, which have a very strong quality media sector, have been able to address these issues successfully, showing high percentages of access to quality paid content and high levels of trust in the news. Analogously, international quality outlets like The Guardian have been able to manage the transition to the digital era successfully,2 emphasising a direct relationship with their readers based on trust and high quality content. In a nutshell, it is our view not only that quality journalism is extremely important in this new age of polarisation and fake news, but also that the same mechanisms that fuel these negative trends at the same time create an opportunity for quality news organisations to thrive, by providing both readers and advertisers with a platform characterised by trusted and high quality news content.

This paper presents a vision for how recent advances in Artificial Intelligence (AI) can support trustworthy high-quality journalism at “every stage of the journalistic value chain” [12]. According to the American Press Institute,3 this value chain involves “the activity of gathering, assessing, creating, and presenting news and information. It is also the product of these activities” (Fig. 1).

Fig. 1. All four stages of news production can be augmented with AI-support for journalists, editors, and newsrooms. Blue arrows show the forward flow of information content from sources towards the audience, whereas red arrows indicate feedback on trustworthiness and other qualities from the audience back to the earlier production stages.

Our vision centres on trustworthiness, which we see as a central challenge for newsrooms today, with the broader goal to contribute to the AI for Goodendeavour [13], through innovative media technologies that help journalists and editors promote informed and engaged citizenship. We emphasise that trustworthy AI-supported journalism requires a fine balance between AI-augmented human tasks and AI-automated routine tasks, acknowledging that “technology is not simply an external tool journalists are forced to assimilate into newswork (though the drive for technological innovation is often fuelled by funders and external actors) but is instead a tool shaped by journalistic practices, needs and norms that similarly alters, sometimes dramatically, everyday newswork” .

In , the gathering and assessing activities have to do with the trustworthiness of the journalistic sources and of the information they provide, whereas creatingand presenting relate to the users’ actual and perceived trust in the news stories they receive. Hence, a vision for trustworthy journalism through AI must take into account both the producers (journalists, editors, newsrooms) and consumers (audience) of news. It must take into account both the regular audience and the adversaries who seek to exploit or diminish trust in the media [14].

A number of contributions have already discussed the relationship between AI and journalism, many of them addressing specific elements of this relationship. For example, [15] discusses the robotisation of journalism, while [16] focuses on the use of AI in investigative journalism. Galily [17] reflects on the possible development of automated journalism using sports journalism as the example, whereas Lewis et al. [18] discuss responsibility for libellous content produced by automated journalism. Finally, Beckett  presents a survey of journalists’ views on how AI can impact on journalistic practices, carried out by JournalismAI,4 “a global initiative that aims to inform media organisations about the potential offered by AI-powered technologies”. According to Lin and Lewis [19], the “one essential thing” that journalistic AI might do for democracy is to “provide accurate, accessible, diverse, relevant, and timely news about public affairs”. Compared to these efforts, our paper focuses more specifically on trustworthiness and on how AI can be harnessed to achieve it.