Journalism's AI Threat and Opportunity
2023 could be the single most disruptive year for news, media, and publishing companies. Almost nobody is prepared.
Hi all,
For the month of January, I am planning on writing a new Substack post each week about technology that is poised to dramatically change the news, media, and publishing industries.
I hope you enjoy,
Matt
It almost goes without saying that the news industry has failed to recognize and act on disruptive changes in the digital age. Whether it has been giving up their dominance over advertising, failing to understand the power of networks, sleeping on mobile-first, or ceding readership relationships to Google, Facebook, etc.
2023 is likely to be the year that two major technologies make the leap out of the lab and into the marketplace in truly disruptive ways. Those are AI and AR.
This week, let’s talk about the threat from AI.
Five years ago I spent a year at Harvard and MIT as a Nieman Fellow, studying how AI & automation might impact the future of news, media, and publishing. My main takeaway at the time was that the media industry was far too focused on short-term ideas around how structured data could be turned into written stories- think turning the box scores from a Red Sox game into a game summary people could read, and not preparing for the day when computers could read and write at a reasonably intelligible level.
In the short five years since my time in Cambridge, computers reading and writing has become an undeniable reality, and 2023 will be the year when this technical breakthrough transitions from a party trick (think ChatGPT) to real products at scale.
How is this a threat to publishers?
Search will be the first big space where this gains traction. Imagine someone types “Hurricane Ian” into search, and instead of getting a search results page full of links to news stories about Hurricane Ian, they get an automatically written summary based on tens or hundreds of stories about Hurricane Ian that an AI system has already read and understood. To top it off, the summary is customized by location, device, reading history, etc. the technology to do this exists today. Will these summaries be perfect? No, but they’ll be fast enough and good enough to complete the customer’s “job to be done” and be economically advantageous to platforms as computing costs continue to decline.
The Search startup Neeva has already launched a “baby” version of this where Neeva provides AI-driven summaries of news articles, saving casual readers a click, and most readers casual readers.
While this feels like a trivial example, it represents a massive shift in the economics of publishing. Search traffic to publishers could plummet. Google sends about 24 billion clicks to news sites each month. It is exceedingly likely that a portion of those clicks to publishers will be supplanted by views of on-platform AI-written summaries.
The argument could even be made by platforms that by having their machines read tens or hundreds of articles from various sources on a politically hot topic, it could generate a more neutral and less biased version.
The impact AI will have on publishing should no longer be a thought exercise. The technology is here and it works. The economic incentives for tech platforms to implement are too strong to ignore.
Publishers need to brace for this change and strategize their moves.
Next week I’ll share more about what I think news organizations can do to take advantage of this new AI technology and better serve both their readers, and their business.