When it comes to thinking about the impact of AI on the media, few are as immersed in this space as Professor Charlie Beckett. “I know more about this than I have known about anything in my life,” he told us when we met in his office last week.
The former news editor, now a Professor at LSE (London School of Economics) has been working almost exclusively in this space since 2019, when he authored a report on artificial intelligence and associated technologies, based on a survey of 71 news organizations in 32 different countries. The latest iteration of this global survey will be published in Fall.
Prior to joining LSE in 2006 as the Founding Director of its journalism think Polis, Beckett was a program editor for Channel 4 News in the UK, responsible for coverage of major stories such as 9/11 and the 7/7 London bombings. And before that, he spent a decade as a senior producer and program editor at BBC News and Current Affairs.
Support from Google News Initiative, under the auspices of the JournalismAI project which Beckett directs, has enabled him and his team to produce a range of training resources, including a JournalismAI Starter Pack, a monthly newsletter and two online courses on machine learning. They have also hosted an Academy for Small Newsrooms and an annual Fellowship designed to foster collaboration and innovation among journalists and technologists around the world.
As Beckett explains, many of these efforts have been bottom-up, informed by the needs of industry leaders and practitioners. Based on this engagement, and Beckett’s ongoing conversations with newsrooms and senior media leaders, here’s a look at the key issues pertaining to AI – and specifically Generative AI – today.
Challenges for media companies
Echoing some of the findings that we reported on from the recent FIPP World Media Congress, Beckett stresses the potential impact of Generative AI on IP and user habits.
“Publishers are worried about things like copyright, data ownership, trying to come to deals with the tech companies… and they’re very worried, obviously, about the disintermediation problem,” he told us.
Acknowledging that disintermediation – the reduction in the use of intermediaries (i.e., publishers) in the content supply chain – is “obviously further down the line,” he nonetheless acknowledged that its impact will be “significant for the publishers.”
This was a theme that Beckett returned to at multiple times during our conversation. The ability of Generative AI tools like Bard to scrape content from a myriad of sources means that ”in theory, you may never need to look at a news website again, because all that stuff has been repurposed by ChatGPT [et al].”
That’s obviously a huge challenge when so many media business models are predicated on attention and serving ads on their own properties. The difficulties that many media companies have faced as a result of aggregators, search and social media – where many outlets have struggled to unlock the monetary value of their content when it was discovered through these channels – look set to repeat themselves.
Alongside this, Beckett shared that media companies are concerned about not having the technological expertise or resources they need to navigate this revolution. This means that many businesses will be “dependent on what’s out there,” using existing tools rather than creating their own. Given how quickly this technology is developing media companies are “very worried that it [publisher-relevant solutions] isn’t gonna happen quickly enough,” Beckett said.
Building off many existing using of AI in media companies, this will include incorporating generative AI into workflows (such as your CMS. It also likely means striking licensing deals similar to the one last month between ChatGPT-maker OpenAI and The Associated Press, to permit legitimate use of publisher content.
A look at the innovation projects being explored by LSE’s current JournalismAI Fellows also provides a valuable insight into the issues that media leaders are grappling with, and how AI might help to solve them. This includes automating visuals, identifying misinformation and fostering a deeper dialogue with audiences.
Opportunities for content creators
Certainly, there are very real challenges around monetization and implementation. However, this is accompanied (at least in some quarters) by an enthusiasm for the potential that Generative AI could unlock.
“There’s no doubt about it,” said Beckett, “publishers just see this as a potential godsend.” The rationale for this stems from a sense that Generative AI can help to reduce costs, while at the same time enabling outlets to generate more content and a broader range of material.
“That’s not necessarily the case,” Beckett cautions, noting that alongside efficiencies, new roles and workflows will also emerge as a result of this technology. However, for some organizations, Generative AI has the potential to be a real boon.
Local news outlets are a good example of this, Beckett says. He highlights Newsquest, one of the UK’s largest regional publishers and part of the wider Gannett family, as one player who is enthusiastic about this AI-led future.
“They’re so excited because they have basically got dozens of local titles that don’t do any local news because they sacked so many people because they were chasing dividends to their shareholders,” Beckett says.
For a company like Newsquest, Generative AI can help to automate a lot more content, freeing up reporters to actually chase down and produce stories. “They see it as a kind of saving grace.”
This argument about liberating resources to invest in deeper work is not new. Nevertheless, the range of output that AI can now produce has grown massively and is only gathering speed. “In a year’s time, people will be using Generative AI relatively routinely in journalism,” Beckett predicts.
Freeing up resources to do innovative, creative work will be an important differentiator and a key means to demonstrate value in this brave new digital world.
“In a world of AI-driven journalism, your human-driven journalism will be the standout,” Beckett forecasts. “You’ve got better journalists, cleverer journalists, more creative journalists who do real reporting… that will be where you stand out.”
Where do we go from here?
AI has been used for a range of functions – from automating earnings reports to delivering dynamic paywalls and content recommendations – for longer than you might realize. Beckett posits that Generative AI is part of this continuum.
At the same time, the breadth of its uses, and the speed with which it is developing, can be daunting. Beckett cautions that, as is with many new tools and technologies, people tend to get over-excited about short-term possibilities and fail to spot the long-term consequences.
One difficulty, of course, is that it is very hard to accurately predict what these impacts may be down the road. Nonetheless, Beckett sees potential implications for the labor force, the role of institutions, and content strategies.
When it comes to employment, Generative AI will lead to rationalization. “Some specific jobs will go,” he says, predicting that “the kind of bland [i.e. formulaic and often entry level] journalism jobs will go completely… And of course, you can either take the money, or you can redirect it.”
Beckett advocates that there’s a business incentive to reinvest these savings. Alongside the need for distinctive content that differentiates your offering in an AI world, he also believes that in a world where unattributable scraped content may be de rigueur, some audiences will gravitate towards trusted media institutions.
We saw this during COVID, he reminds us, with the need for reliable information and concerns about misinformation driving consumers to both local and national/institutional players.
“People, as we saw in the pandemic, will turn to mainstream media when they want,” he observes. “They want somebody who can just reliably, reassuringly, narrate the world… in a way that I can access and find understandable.”
By the same token, he suggests that there will also be a place in this AI world for experts with deep knowledge of nuanced subjects, which can be hard for machines to replicate. Other outlets which speak to your identity and worldview, may also continue to flourish.
Alongside this, the ability to repurpose content for different platforms – and potentially in different formats and tones of voice – may also become the cornerstone of many content strategies. Beckett spoke of our differing media habits and needs during the day, and how AI can respond to this by tailoring content accordingly. “All news is data,” he says. And AI means that it potentially be “reformatted in any way that you wish.”
That will take resources but could help to tackle issues of news avoidance among both news aficionados and those less likely to consume the news. AI may be able to help tackle some of the drivers for news avoidance such as tone, relevance, inaccessibility of reporting, as well as the sense of being just overwhelmed by the sheer volume of content out there.
“You’re optimizing for the problem of overabundance,” he says, “people feel overwhelmed by the amount of choice.” He already sees a shift happening, in which “news organizations are doing fewer stories, but they’re doing them better, and they’re doing them in a more engaging way.” That’s a trend that Beckett sees will only continue in a disintermediated media landscape where the emphasis on quality and distinctiveness will be more important than ever.
Undoubtedly, Generative AI holds the potential to revolutionize the way content is created, distributed, and consumed. Yet at the same time, it also presents some interesting opportunities for media companies. This technology offers unique opportunities to engage audiences in novel and distinct manners, while also fostering trust, relevance, and value in an increasingly noisy digital arena.
As AI continues to permeate media and journalism, its likely impact remains uncertain and challenging to forecast. Nevertheless, its potential to both disrupt and revolutionize the industry is evident. Subsequently, it is vital that media companies understand the associated risks and possibilities. Only by doing this, can they proactively seek to mitigate these threats and move forward using Generative AI as effectively and strategically as possible.