The Future of Journalism is at Risk

Rather than destroying print journalism, AI firms should save it.

Last September I wrote a post arguing “[A]lthough AI is going to change the newsroom, human journalists aren’t going anywhere anytime fast. The reason: humans are irreplaceable in several of the key processes necessary to produce quality, fact-based journalism.”

Although I was right in a general sense — we need human reporting to have a healthy journalistic ecosystem — I was wrong in the sense that human journalists “aren’t going anywhere.” Unfortunately, they will be going somewhere — to different professions — based on the fact that generative AI companies are currently eating the news business, particularly advertising-driven text-based journalism. It is hard to imagine how publishers will be compensated unless there’s a large scale intervention or AI company executives come to their senses.

The advent of Perplexity, the latest LLM unicorn, was the first alarm bell in this regard. Perplexity aggregates information to provide summaries of informational topics — a query of “what’s going on in Gaza” will return multiple AI-generated paragraphs citing news publisher sources. On the New York Times Hard Fork podcast, Perplexity’s CEO admits that he had no immediate plans to compensate publishers for incorporating their content into their AI summaries, which in effect means that he expects publishers to give over their content for free. The intent of the summaries is that users will not have to click through to read the source material. A little clickable CNN icon in the corner of a Perplexity summary is not going to lead to substantial click-throughs and ad revenue for CNN.

Similarly, Google Search is beginning to add generative AI summaries to the top of the results page. Ads-driven outlets that were dependent on traffic from Google Search will no longer receive it, as users come to trust Google to “do the Googling for them.” Again, I find it hard to imagine that users will want to click through if most of them get all the information they need from the news summary.

The news summaries debuting in Google Search.

Moreover, the New York Times lawsuit against OpenAI could prove a bellwether for the future of the news industry. If using paywalled journalism to train algorithms is considered “fair use,” it logically follows that no content is safe from algorithmic extraction. Rather than taking the legal approach, large publishers including News Corp, The Atlantic, and Vox Media are cutting deals, trying to find revenue while they still have the chance.

All of these trends mean that in just a few years, the ads-based model for journalism will mostly dry up, which means journalists will be compensated even less fairly for the quality reporting they produce. People will get their information from summaries whose producers will many times not be able to tell you from which sources the information was extracted. AI firms will have all the power to determine how those summaries appear and how much credit to give to the publisher content that helped build them. Large corporations will have power to tell readers exactly what to think — gone will be the days where readers scan the web for multiple perspectives on the issues that matter.

And, when algorithm companies eat the print journalism industry, there will be no one left to do shoe leather reporting and hold truth to account. Apart from large philanthropists and the small group of people willing to pay for news, there will be few people interested in paying them.

If you think I’m painting a bleak picture, you would most certainly be right. However, there is a kernel of hope: destroying the news business will cut off much of the content AI companies need to provide accurate information and stay current with world events.

When taken to that extreme, AI companies are aligned with media organizations - new original content means consistently relevant algorithmic output. If we accept that alignment as a given, there are win-win deals available between media companies and algorithm providers that shore up publishers while providing data for algorithms. Although eating the news business may be in the short term interest of LLM firms, a world without journalism means the quality of these algorithms will decline in the long run.

The way journalism is financed may be in the process of changing. Perhaps journalism will become an entirely philanthropic enterprise in the long run. Perhaps the information gatherers will be funded by the aggregators (the LLM companies). On the hopeful side, perhaps, even in the short term, LLM providers will realize their alignment with news companies and will cut deals to help publishers earn revenue for the increased distribution AI algorithms will provide. One such solution could be providing more prominent publisher attribution in summaries or cutting deals to direct results to specific trustworthy publishers based on the topic of the query.

However, a world without journalism is a world more hostile to democracy and healthy public discourse — and a world that’s bad for AI firms. Leaders in those firms need to get more creative about how to save journalism rather than simply eating it for lunch.