Are Journalists Junk?

AI writing newspapers — the rise of the machines continues

The full implications of artificial intelligence (AI) for the writing community became clearer this week with the revelation that the company owning the “Daily Mirror” and the “Express” has already published its first articles written using AI.

Like the print workers forty years ago, journalists are the latest employees in the newspaper industry to feel the impact of the march of technology. Older readers may recall the bitter battles in 1986 between police and strikers on the picket lines in Wapping, East London, when print workers began losing their jobs to the new word-processing technology that we now all take for granted.

The same process of technology replacing human skills appears about to begin at scale for journalists. Of course the arrival of AI at speed and scale is causing ripples of concern and alarm in numerous other sectors like education and finance, and there are even fears that criminals are using AI to refine their frauds, by improving the linguistic complexity of phishing attempts.

AI writing newspapers

According to Jim Mullen, the Chief Executive of “Reach” (formerly the Trinity Mirror group), the company let a bot produce three articles last week after a working group explored the possibilities for the use of AI. The articles were first published on a local news site, “InYourArea.co.uk”, one of which was a “listicle” (an awful word meaning mixture of a list and an article) titled “Seven Things to do in Newport”, though it is a credit to the technology that it managed to find seven. AI seems particularly good at producing numbered lists, which are one of the main signs that it has been involved in writing an article, along with a bland, repetitive style, and American spellings less familiar to those of us in the UK.

Sources at the publisher claim that news of the AI test had put some journalists “on edge” as Reach continues to focus on heavy cost-cutting. The loss of 200 roles announced in January seems just the tip of the iceberg, as soaring inflation and newsprint prices reaching a 15-year high have taken their toll. This comes after years of decline in the readership of print newspapers, as more content is viewed online, and there is also a trend away from newspaper readership generally, and consuming news in other ways, if at all.

No hidden agenda?

Despite the journalists’ concerns, Mullen assured the journalists that the development of AI is not part of a “hidden agenda” to ultimately make big savings by being able to cut human staff. Given that the company’s 4,000 employees represent its biggest cost, the concern is perhaps founded. After all not all the journalists will be able to retrain as restaurant reviewers, and Microsoft will probably already be working on that ability in their next version update.

Mullen insists that the AI-produced articles were led by editorial, and that the data was put together by a journalist. He claims that whether it was good enough to publish was decided by an editor. In areas based on data, such as “things to do,” weather, and “what’s local traffic like?” pieces, he argues that AI might be reliable enough to take on the production of content. One can imagine some journalists might welcome AI covering the more tedious reporting such as council meetings.

However, the rapid rise of Microsoft’s “ChatGPT” and Google’s hurriedly released potential challenger “Bard”, has resulted in publishers focusing on the potential for using machine learning in the wider journalism production process. The likes of “BuzzFeed” and “Associated Press” have already started working with AI to help produce their quizzes and articles, and this practice is likely to spread rapidly, though to be fair, Google News has been using a similar process for many years in automatically selecting certain stories for their newsfeed.

In his book “Breaking News — the remaking of journalism and why it matters now”, former Guardian editor Alan Rusbridger speaks of the arrival of the Google News service:

“In October 2002 one or two colleagues stumbled across a new service called Google News. One self-mockingly responded by slumping across his desk in a pose that suggested there was little point in carrying on. How could one compete with this all-seeing eye on the world, hoovering up anything that happens, anywhere on the planet and alerting users within minutes in a remorseless perpetuum mobile of breaking information? The most sinister thing from the point of view of the average desk editor sitting in Farringdon Road or Wapping or Canary Wharf was the little line at the bottom of the site: the selection and placement of stories on this page were determined automatically by a computer programme.”

As Reach continues to struggle with a fall in digital income, the company is investing heavily in breaking into the US market. Despite promising a 5%-6% cut in its cost base this year, shares in Reach fell by almost 13% on Tuesday as investors expressed concern over the tough trading conditions.

Complex ethical issues

The use of AI in journalism is a complex issue that raises ethical concerns. As machines become more advanced, journalists may soon find themselves being replaced by bots altogether. While AI may be reliable in certain areas based purely on data, it is important to remember that journalism is not just about making numbered lists of things to do in Newport, or simply presenting regurgitated facts and figures — it is about telling stories that engage and inform readers, in a lively and readable style that is rarely matched by current AI output, that is frequently bland and repetitive.

The unique role of journalists to witness and tell the truth

Journalists often write opinion pieces on controversial subjects that require empathy or emotion to convey them well, or first-hand accounts of events they have experienced. Take for example the extraordinary reporting of the bombing of Guernica during the Spanish Civil War by Times journalist George Steer, and captured below in the famous painting by Picasso -

Or the fact that the founding of “The Guardian” newspaper itself arose in response to the 1819 Peterloo massacre in Manchester, when peaceful protestors daring to demand universal enfranchisement were shamefully mown down in their dozens by troops on horseback. Or more recently the brave reporting from Kiev by Clive Myrie as the first Russian bombs fell, and the excellent reporting of the Turkey/Syria earthquake by BBC journalist Anna Foster who just happened to be in the right place at the right time to report on the tragedy.

Humans competing with machines

As we move forward, we must ensure that we do not sacrifice the quality and existence of the noble profession of journalism and thousands of jobs, on the altar of cost-cutting and efficiency.

The rapid introduction of AI without debate, consultation or regulation, also raises huge questions about what this level of automation and technological interference means for the very future of what it means to be human. Otherwise the technology will devalue all of us, and real people will find themselves competing with machines to be able to earn enough to support their families.

Bot-spotting difficulty

Illustrating the problems that will come in trying to regulate AI, a new study by the web tool reviewing company “Tooltester” has found that more than half of the public are apparently unable to differentiate between words penned by AI chatbots like “ChatGPT” and those created by human writers, with the younger cohort of Generation Z being the most challenged.

The researchers discovered that a staggering 53% of participants under 65 could not distinguish between content produced by humans, AI, or AI combined with human editing.

In contrast, individuals aged 65 and over were able to correctly spot AI content over 50% of the time. The study involved 1,900 Americans, who were asked to identify whether writing was authored by a human or an AI in diverse sectors, including health and technology.

AI content warning labels wanted

The research revealed that people want warning labels to indicate when AI is used to generate content. Over 80% of respondents believed that firms publishing blogs or news articles should give a caution if AI was employed, with 71% indicating they would trust a company less if they used AI-generated content without being transparent about it.

Robert Brandl, CEO and Founder of Tooltester, who conducted the survey, stated that the findings implied that people might have to depend on artificial intelligence disclosures online to ascertain what is human or AI-generated since people cannot tell the difference.

Robert Brandl states:

‘The fact that younger readers were less savvy at identifying AI content was surprising to us. It could be suggestive of older readers being more cynical of AI content currently, especially with it being such a big thing in the news lately. Older readers will have a broader knowledge base to draw from and can compare internal perceptions of how a question should be answered better than a younger person, simply because they have had many more years of exposure to such information. One study from the University of Florida showed that in fact younger audiences are as susceptible to fake news online as older generations, therefore, being young and potentially more tech-savvy is not a defense to being tricked by online content.’

Familiarity breeds contempt

The study highlighted that familiarity with “generative AI” like ChatGPT appeared to help, as only 40.8% of people were unfamiliar with the technology, and they were more likely to identify AI content correctly. Nonetheless, Brandl cautioned that AI writing is far from perfect and can contain numerous inaccuracies. Some respondents presumed any text, whether written by humans or AI, was AI-generated, which could reflect the cynical approach many people are taking to online content.

The ability of people to identify AI-generated content varied by sector, with AI-generated health content being the most deceptive, with 56.1% of individuals wrongly believing that a human or a combination of human and AI edited it. On the other hand, technology was the area where people could most quickly recognize AI-generated content, with 51% correctly identifying it.

AI aiding criminals

Cybersecurity researchers have also warned of an impending wave of AI-written phishing attacks and fraud. Norton, a cybersecurity company, cautioned that cybercriminals are employing AI tools like ChatGPT to create ever more efficient “lures” to rob victims, by refining linguistic complexity.

A spokesperson for Darkspace, another cybersecurity firm comments:

“Darktrace has found that while the number of email attacks across its own customer base remained steady since ChatGPT’s release, those that rely on tricking victims into clicking malicious links have declined while linguistic complexity, including text volume, punctuation and sentence length among others, have increased…..This indicates that cybercriminals may be redirecting their focus to crafting more sophisticated social engineering scams that exploit user trust……ChatGPT has [not] yet lowered barriers to entry for threat actors significantly, but it does believe that it may have helped increase the sophistication of phishing emails, enabling adversaries to create more targeted, personalised, and ultimately, successful attacks.”

Press the pause button

As AI chatbot ChatGPT now has an estimated 13 million daily users, it is currently the fastest-growing application in the history of the internet, according to analytics firm SimilarWeb. The OpenAI-backed private firm, Microsoft Corp, made ChatGPT accessible to the public for free in late November. In the space of a few months other companies such as Google are now locked in an arms race to make their products ever more efficient.

It is clear that whether in journalism or other sectors, the rapid introduction of AI without debate, consultation or regulation, needs urgent action to deal with some of the issues it is causing. The pace of technological advance needs to matched in pace by regulation and a process to give the public a say in how it is introduced, and who reaps the financial rewards.

To join a new Facebook group “Medium Matters” where writers can support each other and share articles.

Mastodon- you can find me here