The rise of AI in content production may disrupt the relationship between quality and its competitive edge. “Those repurposing content, not those creating it, could reap the benefits,” warned Professor Dariusz Jemielniak in an interview with INMA.
By the end of June 2023, U.S. news rating service NewsGuard identified 301 Web sites that operated with little or no human oversight and published articles written largely or entirely by AI bots. The number of AI-generated news Web sites increased six-fold since May, when NewsGuard identified 49.
Many of those Web sites appeared to be entirely financed by programmatic advertising, and some churned out huge volumes of articles — one such site produced on average 1,200 articles a day! For comparison, France’s Le Monde publishes about 100 articles a day, The New York Times, up to 250 articles a day. U.K.-based Mail Online, 1,700 articles per day.
This is just the beginning. “Flooding the market with cheap information, AI can lead to decrease in overall quality of the Web and misinformation,” said Jemielniak, who studies an intersection of management and the Internet for Harvard and Poland’s Kozminski University.
Together with Professor Aleksandra Przegalinska, he authored a new book, Strategising AI in Business and Education, published in 2023 with Cambridge University Press.
The fourth wave of disruption
In Jemielniak’s opinion, AI represents the “fourth wave of disruption” to the news media industry, following the digital upheaval brought by news portals, search engines, and the prosumer revolution on social media.
With AI drastically driving down production costs, businesses relying today on low-cost content might lose their cost advantage while struggling to differentiate.
“Off-the-shelf AI tools will revolutionise the production model for all,” explained Jemielniak in the interview with INMA. “But they won’t provide a competitive edge for anybody. We may see thousands of generic Web sites for which the only difference will be branding.
“This is a major disruption, a potential tsunami, that actually our current ecosystem doesn’t really have a good answer to,” said Jemielniak, who also complained that AI companies were training algorithms on original content from media sites without attribution, nor licensing.
He knows the problem first-hand as a trustee of Wikimedia Foundation. Its popular online encyclopaedia Wikipedia is routinely scrapped by AI companies.
Human edge over AI
I asked for possible scenarios. Apart from the negative developments, Jemielniak imagined the positive ones, too: “AI tsunami might soon enough cause users to get frustrated by low-quality content and put a premium on journalism, including information verification and investigative reporting.”
In such an optimistic scenario, reputable publishers would survive the AI tsunami due to their brand trust built over years and established quality control processes.
“AI cannot really send anybody to witness events. It cannot gain trust of sources revealing hidden information for investigations. It doesn’t question the official sources. These journalistic processes are difficult to substitute with AI,” believed the academic.
News customised with AI
Professor Jemielniak anticipated a new dawn for personalised content and advertising: “Generative AI is the marketing Holy Grail.”
While the backbone of journalism is likely to remain human-based, the delivery might be increasingly handled by machines. As generative AI gets better, Jemielniak envisions full news customisation based on a reader’s language and preferences: "The cost of delivering news tailored specifically for you is now negligible. My guess will be that in the end, we all will be getting fully customised news written in the style and language that readers like to read.
“Today, many media outlets follow a mimicry strategy, copying what others are doing, to maximise its content’s appeal and reduce the risk of investment in niches. AI may push the media away from this paradigm to increase differentiation through personalisation to many narrow segments. Once the paradigm shifts, everybody might follow.”
This idea is described in Jemielniak’s book as the likely shift from two classic strategies positioning the business on low cost or differentiation to serving multiple niches at scale.
Investing in AI
Not developing AI competencies may lead to a business’ downfall, as others could outperform in terms of cost, speed, and quality. “Developing an AI strategy, starting with modest goals, and creating an AI-conscious business culture is crucial,” urged the professor.
The AI implementation process should be iterative, with many goal adjustments and experiments along the way. Whether AI is used to increase effectiveness, efficiency, or both, it’s critical to prioritise accordingly.
While large companies may build their own AI models, small- and mid-sized will likely rely on off-the-shelf solutions from vendors. Although these are already widely available and easy to use, they provide no competitive edge.
As AI technology matures and becomes widely accessible, competition will likely shift to areas like marketing and branding. Developing unique AI solutions may become less important than the ability to brand and market existing solutions effectively.
In the meantime, measuring ROI from AI implementations can be challenging due to the significant upfront investment and potential for zero outcome. “It’s important to set clear goals and directives and align data scientists with business teams and management to avoid misalignment,” he recommended.
Training for work with AI
Jemielniak advised aggressive training for employees on AI-driven tools to help people prepare and adapt to working alongside machines: “Some jobs will inevitably disappear, but there will always be a need for professional skills. For example, while everyone can now take photos, professional photographers are still in demand due to their unique skills.”
How to survive generative AI-driven search? Read my original analysis.