Over 70 percent of senior editors and CEOs of media organisations believe that AI and Generative AI will lower the public's overall trust in the news, according to a report published by the Reuters Institute for the Study of Journalism on Tuesday, 9 January.
The annual report titled 'Journalism, Media, and Technology Trends and Predictions 2024' surveyed over 314 CEOs, editors-in-chief, and heads of digital or innovation at print, broadcasting, and digital news organisations from 56 countries, including India.
The report also reveals that up to 48 percent of news executives think that there would be very little money for news publishers in any licensing deal with AI platforms.
Furthermore, 35 percent of them believe that most of the money (from such licensing deals) would go to big publishers.
News publishers' apparent distrust of the AI industry comes on the heels of a copyright infringement lawsuit filed by The New York Times (NYT) against OpenAI, the company behind popular AI chatbot ChatGPT.
While NYT is far from the first to sue OpenAI, it sheds light on the major copyright battle brewing between developers of generative AI models and those who own the rights to the data that's used to train such AI models.
In the midst of these fundamental legal issues, how is AI currently being used in newsrooms across the world? What are the ways this technology could disrupt the media industry? And what else are news executives worried about in the new year? Take a look at the key findings of the Reuters Institute report.
AI in Newsrooms: What Are Media Bosses Worried About Most in 2024?
1. How Are Newsrooms Using AI?
The most common use of AI tools in newsrooms is for back-end automation tasks such as tagging, transcribing, and copy-editing, according to the Reuters Institute report.
Over 56 percent of news executives said that the use of AI for such back-end automation tasks was a high priority.
Two years ago, only 29 percent of news executives had found this use case to be high priority, indicating that "newsrooms have been coming to terms with the implications of advances in AI – and generative AI in particular."
According to the report, other common applications of AI in newsrooms are:
Recommender systems (37 percent)
Creation of content with human oversight (28 percent)
Commercial uses (27 percent)
Coding and product development (25 percent), where publishers say they have seen very large productivity gains.
News-gathering (22 percent), where AI may be used to support investigations or in fact-checking and verification.
The report also said that various news organisations have been trying out AI tools for the following tasks:
Summarisation: A number of European publications have been adding AI-generated bullet points at the beginning of their articles. They found that these summary bullet points increased their overall engagement, as per the report.
Headline testing: "A number of publications have been experimenting with AI-driven headlines optimised for search, which are then checked by editors," the report read.
"The curation and selection of the headlines and links that consumers see on the website is now also automated, leading to a significant boost in click-though rates."
German tabloid Express.de CEO Thomas Schultz-Homberg.Copy-editing, note-taking, and transcription: Using AI tools to assist in these tasks is becoming more widespread in newsrooms, as per the report. Transcription tools in small languages have also significantly improved in the last year, it added.
However, AI aiding in these processes could mean layoffs in the news media industry. “The functions of editor-in-chief, editors, proofreaders, secretaries, and photo editors will no longer exist as they do today," publisher Axel Springer SE had said in June last year.
Translation: The report points out that French daily Le Monde uses AI to assist in the "initial translation" of stories from French to English – albeit with several human checks. "The software has been customised to recognise Le Monde’s style book and spellings," it added.
Image/article generation: Generating images using tools like Midjourney is one of the most popular uses of AI by news organisations. A few tabloid sites are also reportedly using LLM models to generate articles on various subjects.
"Human editors still decide which stories get covered and review every piece of content, but the writing, structure, and tone are effectively outsourced to AI," the report read.
AI presenters and news-readers: Several radio stations are using AI voice clones to anchor the night shift "complete with commentary about the music."
"Two radio stations in the west of England use a synthetic voice to turn text into an hourly radio bulletin. The company says they wouldn’t have been able to afford to employ a human journalist and that the voice is so realistic, few people have been able to tell the difference," the report stated.
Expand2. What Are the Risks of Using AI in Newsrooms?
"Publishers are ambivalent about using AI for content creation. More than half of our respondents (53 percent) consider it the biggest reputational risk," the report read.
Additionally, over 33 percent of respondents pointed out that using AI for news-gathering purposes was also a risk. "By contrast, back-end automation (9 percent), distribution, and coding are considered lower risk," the report added.
For context, here are some news publishers whose experiments with AI went seriously wrong:
CNET published articles that were written using an AI tool in 2023. However, the outlet ended up issuing corrections on 41 out of 77 such AI-generated stories, The Verge reported. CNET also received flak for not disclosing, at the outset, that it was publishing AI-generated content.
Sports Illustrated (SI) was also found to have included product reviews from a third party that were alleged to be at least partly written by AI, without proper disclosure.
Gizmodo published its first AI-generated article titled 'A Chronological List of Star Wars Movies & TV Shows, but fans were quick to point out the factual errors in the article.
Men's Journal published an AI-generated article about low testosterone in men, which reportedly contained blatant falsehoods and claims that aren't supported by data.
The Guardian accused Microsoft AI of damaging its reputation after the latter placed a "distasteful" AI-generated poll next to a news report about a woman's murder. The poll reportedly asked readers to speculate on the reason behind the woman's death.
Channel 1 News, based out of Los Angeles, is looking to launch a personalised AI-driven newscast "that learns what stories a viewer wants to see and promises to deliver them in any language."
"Journalists worry about the implications of this kind of automation for jobs, and for trust in the news media," the report read.
Expand3. What Else Troubled Media Execs in 2023?
Over 63 percent of the media leaders surveyed say they are worried about a sharp decline in referral traffic from social media platforms, as per the report published by Reuters Institute.
The report cited figures from data analytics provider Chartbeat and said that in 2023, news publishers saw that:
Facebook traffic fell 48 percent.
X (formerly Twitter) traffic fell 27 percent.
Instagram traffic fell 10 percent.
"Both Facebook and X/Twitter as a place for traffic – or even news consumption – are pretty much dead for news organisations in India," a leading news editor was quoted as saying in the report.
"Interest in video networks such as TikTok (+55) and YouTube (+44) remain strong while Google Discover is becoming a more important but volatile referral source," the report read.
Meta's decision to open up broadcast channels for publishers on platforms such as WhatsApp and Instagram is pushing news organisations to focus more on these platforms, it added.
Expand4. Content Bundling, AI Bots: 5 Key Predictions for 2024
Decline in search traffic: “Reaching audiences online is getting tougher as Facebook pulls back from news and X becomes less welcoming for publishers. The big fear is that search traffic may be next, as AI-powered results provide answers directly in the interface, rather than offering so many links to news sites," Nic Newman, the author of the Reuters Institute report, said.
Google Search, Microsoft Bing, and others have been exploring the integration of AI with their respective search engines so that direct answers are provided to the queries of users, instead of showing them a list of links to websites.
Increased focus on WhatsApp: "Publishers say they will be putting more effort into WhatsApp this year following the launch of functionality that allows a range of personalities and brands to create broadcast channels," the report revealed.
More news-reading apps: Reuters Institute predicted that AI-based apps "that change the language of news to improve relevance and understanding for particular audiences will be an increasing feature of the news landscape in 2024."
"Bots, apps and browser extensions with similar capabilities may spread rapidly in 2024, putting pressure on news organisations to create similar features," it added, offering Artifact (developed by the co-founders of Instagram) as an example.
Bundling of digital news content: "Expect to see a significant shift towards bundling of digital news and non-news content as large publishers look to lock in existing customers. All access subscriptions will include games, podcasts, magazines, books, and even content from other publishers," the report read.
"Big tech platforms will also be leaning further into paid business models as they look to reduce their dependence on advertising. X, Meta, and TikTok will offer more premium services this year including ad-free and privacy-friendly options," as per the report.
Copyright battles, AI arrangements: The report points out that AI industry players may face growing pressure from publishers to compensate for any loss in traffic, training their systems on historical data, or delivering real-time news.
However, it also mentions a few voluntary arrangements in place such as the deal between OpenAI and publisher Axel Springer for content drawn from Bild, Politico and Business Insider. But the money from such deals may not be shared equally, the report warned.
(At The Quint, we question everything. Play an active role in shaping our journalism by becoming a member today.)
Expand
How Are Newsrooms Using AI?
The most common use of AI tools in newsrooms is for back-end automation tasks such as tagging, transcribing, and copy-editing, according to the Reuters Institute report.
Over 56 percent of news executives said that the use of AI for such back-end automation tasks was a high priority.
Two years ago, only 29 percent of news executives had found this use case to be high priority, indicating that "newsrooms have been coming to terms with the implications of advances in AI – and generative AI in particular."
According to the report, other common applications of AI in newsrooms are:
Recommender systems (37 percent)
Creation of content with human oversight (28 percent)
Commercial uses (27 percent)
Coding and product development (25 percent), where publishers say they have seen very large productivity gains.
News-gathering (22 percent), where AI may be used to support investigations or in fact-checking and verification.
The report also said that various news organisations have been trying out AI tools for the following tasks:
Summarisation: A number of European publications have been adding AI-generated bullet points at the beginning of their articles. They found that these summary bullet points increased their overall engagement, as per the report.
Headline testing: "A number of publications have been experimenting with AI-driven headlines optimised for search, which are then checked by editors," the report read.
"The curation and selection of the headlines and links that consumers see on the website is now also automated, leading to a significant boost in click-though rates."German tabloid Express.de CEO Thomas Schultz-Homberg.
Copy-editing, note-taking, and transcription: Using AI tools to assist in these tasks is becoming more widespread in newsrooms, as per the report. Transcription tools in small languages have also significantly improved in the last year, it added.
However, AI aiding in these processes could mean layoffs in the news media industry. “The functions of editor-in-chief, editors, proofreaders, secretaries, and photo editors will no longer exist as they do today," publisher Axel Springer SE had said in June last year.
Translation: The report points out that French daily Le Monde uses AI to assist in the "initial translation" of stories from French to English – albeit with several human checks. "The software has been customised to recognise Le Monde’s style book and spellings," it added.
Image/article generation: Generating images using tools like Midjourney is one of the most popular uses of AI by news organisations. A few tabloid sites are also reportedly using LLM models to generate articles on various subjects.
"Human editors still decide which stories get covered and review every piece of content, but the writing, structure, and tone are effectively outsourced to AI," the report read.
AI presenters and news-readers: Several radio stations are using AI voice clones to anchor the night shift "complete with commentary about the music."
"Two radio stations in the west of England use a synthetic voice to turn text into an hourly radio bulletin. The company says they wouldn’t have been able to afford to employ a human journalist and that the voice is so realistic, few people have been able to tell the difference," the report stated.
What Are the Risks of Using AI in Newsrooms?
"Publishers are ambivalent about using AI for content creation. More than half of our respondents (53 percent) consider it the biggest reputational risk," the report read.
Additionally, over 33 percent of respondents pointed out that using AI for news-gathering purposes was also a risk. "By contrast, back-end automation (9 percent), distribution, and coding are considered lower risk," the report added.
For context, here are some news publishers whose experiments with AI went seriously wrong:
CNET published articles that were written using an AI tool in 2023. However, the outlet ended up issuing corrections on 41 out of 77 such AI-generated stories, The Verge reported. CNET also received flak for not disclosing, at the outset, that it was publishing AI-generated content.
Sports Illustrated (SI) was also found to have included product reviews from a third party that were alleged to be at least partly written by AI, without proper disclosure.
Gizmodo published its first AI-generated article titled 'A Chronological List of Star Wars Movies & TV Shows, but fans were quick to point out the factual errors in the article.
Men's Journal published an AI-generated article about low testosterone in men, which reportedly contained blatant falsehoods and claims that aren't supported by data.
The Guardian accused Microsoft AI of damaging its reputation after the latter placed a "distasteful" AI-generated poll next to a news report about a woman's murder. The poll reportedly asked readers to speculate on the reason behind the woman's death.
Channel 1 News, based out of Los Angeles, is looking to launch a personalised AI-driven newscast "that learns what stories a viewer wants to see and promises to deliver them in any language."
"Journalists worry about the implications of this kind of automation for jobs, and for trust in the news media," the report read.
What Else Troubled Media Execs in 2023?
Over 63 percent of the media leaders surveyed say they are worried about a sharp decline in referral traffic from social media platforms, as per the report published by Reuters Institute.
The report cited figures from data analytics provider Chartbeat and said that in 2023, news publishers saw that:
Facebook traffic fell 48 percent.
X (formerly Twitter) traffic fell 27 percent.
Instagram traffic fell 10 percent.
"Both Facebook and X/Twitter as a place for traffic – or even news consumption – are pretty much dead for news organisations in India," a leading news editor was quoted as saying in the report.
"Interest in video networks such as TikTok (+55) and YouTube (+44) remain strong while Google Discover is becoming a more important but volatile referral source," the report read.
Meta's decision to open up broadcast channels for publishers on platforms such as WhatsApp and Instagram is pushing news organisations to focus more on these platforms, it added.
Content Bundling, AI Bots: 5 Key Predictions for 2024
Decline in search traffic: “Reaching audiences online is getting tougher as Facebook pulls back from news and X becomes less welcoming for publishers. The big fear is that search traffic may be next, as AI-powered results provide answers directly in the interface, rather than offering so many links to news sites," Nic Newman, the author of the Reuters Institute report, said.
Google Search, Microsoft Bing, and others have been exploring the integration of AI with their respective search engines so that direct answers are provided to the queries of users, instead of showing them a list of links to websites.
Increased focus on WhatsApp: "Publishers say they will be putting more effort into WhatsApp this year following the launch of functionality that allows a range of personalities and brands to create broadcast channels," the report revealed.
More news-reading apps: Reuters Institute predicted that AI-based apps "that change the language of news to improve relevance and understanding for particular audiences will be an increasing feature of the news landscape in 2024."
"Bots, apps and browser extensions with similar capabilities may spread rapidly in 2024, putting pressure on news organisations to create similar features," it added, offering Artifact (developed by the co-founders of Instagram) as an example.
Bundling of digital news content: "Expect to see a significant shift towards bundling of digital news and non-news content as large publishers look to lock in existing customers. All access subscriptions will include games, podcasts, magazines, books, and even content from other publishers," the report read.
"Big tech platforms will also be leaning further into paid business models as they look to reduce their dependence on advertising. X, Meta, and TikTok will offer more premium services this year including ad-free and privacy-friendly options," as per the report.
Copyright battles, AI arrangements: The report points out that AI industry players may face growing pressure from publishers to compensate for any loss in traffic, training their systems on historical data, or delivering real-time news.
However, it also mentions a few voluntary arrangements in place such as the deal between OpenAI and publisher Axel Springer for content drawn from Bild, Politico and Business Insider. But the money from such deals may not be shared equally, the report warned.
(At The Quint, we question everything. Play an active role in shaping our journalism by becoming a member today.)