If you wake up on your birthday to a slew of messages from your friends and family. They all contain photos, memories, and well-wishes for you. The catch is, the birthday notes were drafted and sent by an artificial intelligence (AI), and your friends and family didn’t even realise it was your birthday. Are these friends and family maintaining a meaningful relationship?

This was the question thrown to the audience by BBC product manager David Caswell at the end of Extra! Extra! AI & the Future of News, a panel discussion held during the Scottish AI summit on March 28th at the University of Strathclyde in Glasgow. He believes that maintaining the bond of trust should be paramount as newsrooms increasingly adopt AI in their workflows. That part of upholding your brand, much like upholding a friendship, comes from transparency and authenticity in making content. 

An early report on the impact large language models (LLMs) will have on the labour market showed journalists, news analysts, and reporters to be among the highest exposed occupations to AI technologies, with between 82-100% of work being exposed to AI, according to a rubric created by the researchers.  

Many newsrooms across the world are already using AI software to publish content, mostly news events which are easy to distil into quick stories – the Washington Post’s coverage of the Olympic Games, or the BBCs coverage of the 2019 general election being two examples

AI’s ability to assist workflows for journalists, paired with its increased accessibility present’s an existential threat to the profession. However, pundits believe there is an invaluable angle to human-made journalism that won’t be taken over by AI any time soon.  

The myth of objectivity

More nuanced commentary or investigative pieces, which require the journalist to situate events on a larger backdrop will still need a human element for a long time, according to Caswell. The development of AI in journalism therefore subverts the myth of objective journalism that was rampant in the Trump-era media landscape. 

Where access to large followings without rigorous fact checking allowed dis-information campaigns to sway public opinion. Now that there is an objective third party which can break news with only the information fed into it and no external influence. This can be beneficial when accurate information is used by the LLM however, malicious actors may also use false facts to feed the LLM, and create a journalistic piece of equal quality. Therefore the value added by the journalist’s life experience and worldview becomes sought after in the news cycle. 

“There are some special journalists that throughout their career did some kind of work and we trust them, we trust what they say,” said Atoosa Kasirzadeh, panel member and Chancellor’s Fellow at University of Edinburgh philosophy department and Research Lead of the Centre for Technomoral Futures. 

“With (AI journalism) we are going to see the production of loads of new material that sound like coherent texts,” she said. “Unfortunately, we don’t have good tests to distinguish whether the content that is generated is really factual and the reason for that is in many areas, or many domains of discourse or conversation, it’s just hard to define.” 

Kasirzadeh identifies here the grey area of the objectivity debate around journalism more generally, which is that the truth is impossible to define. For dissemination of difficult areas of life, consumers rely on journalists with a track record of honest and insightful work. Large media conglomerates work hard to build reputations that stand for ethics and transparency in their process over the myth of objectivity. 

While truth or objectivity in journalism is difficult to define in AI models, disinformation remains an even bigger threat to the media landscape. Kasirzadeh points out that malicious actors can use technologies like AI to make information quickly and at a low cost. 

“We are going to see an explosion of information that is going to confuse many of us about what is fact and what is fiction,” said Kasirzadeh. “It’s very important for the responsible journalists community to think about how we are going to challenge and address these kinds of challenges.”

Recommended

• UK Gov Launches First AI White Paper Amid Open Letter Pushback
• Scot-Secure | Just One Day to Go!
• More than Just Code: Soft Skills Prove Vital to Developers

Verification and transparency

This shift towards verifying fact from fiction is a different beast from the “fake news” era, which was an effect of social media’s growing popularity.

As AI takes a larger role in publishing and promoting content on our social media feeds, AI and Media professor at the University of Surrey Bahareh Heravi says “we are entering a new era of literacy, information literacy, AI literacy, where we need to learn how to deal with this amount of automated generated content, we need to increase our critical awareness, critical thinking to be able to kind of criticise whatever we see, even more than before.”

Heravi identified three categories for use of AI in journalism: Discovery of news, assistance in production, and optimised publishing. This is because many of these workflows are formulaic in nature, allowing AI to easily learn the best way to complete the task. 

Meaning AI’s best use is to assist journalists in their workflows, which would be an existential threat to some in the industry, but also improve the quality and reach of the overall product. What all panel members agreed on is the use of AI in news must be clearly cited as being the author or gatherer of data.