I have an AI prediction of my own. I give it a 10% likelihood that AI will destroy all media jobs within the next decade.
That’s if it doesn’t kill all humanity first. But it too just might be averted by international cooperation.
I sensed a shift in how AI is giving publishers and editors cause for concern at this year’s Future of Media Technology Conference in London. They have moved on from the worries of a few years ago, about the impact of generative AI on the supply side being quicker than journalists at so many tasks, albeit sometimes less accurate with it. AI as a journalism or publishing tool is now becoming routine for editors. It’s still developing but we’re getting used to working with it for various tasks.
The bigger worry now is about AI’s impact on the audience demand and engagement side, from content discovery to subscriptions. The threat, more specifically, from AI search and from agentic AI.
AI search is hitting publishers hard – some really hard.
Search algorithms have always evolved, usually in ways that hurt online publishers. AI search is another such evolution, and a step change if not necessarily a cliff edge. In short, too many eyeballs are now just scanning the AI summaries in Google, getting enough of what they want from those and not bothering to click the few source links that are sometimes now provided.
This matters more to publishers, whose content is their product, than marketers, who just need to get their message across.
Publishers are left with the breadcrumbs of the odd citation, but they can’t be relied upon either. AI search is still a bit of a ‘wild west’: Your site’s position in search results can vary up and down wildly, from hour to hour. And while Google skews more to evergreen content and ChatGPT more to live and timely news, publishers can’t jump on either direction right now because things will evolve. Few users are clicking through anyway.
So, do publishers block and / or license access for the AI bots? Those already locking down their content to human subscribers may be in a better position to charge the bots for scraping their content. But it’s not a silver bullet.
Charging the bots
Locking out the bots could present further problems in that publishers may not want to lock them out entirely. The conference heard there could be opportunities in charging the robots to scrape your content.
News Corp says it’s following a ‘woo and sue’ strategy. The value in deals with AI platforms is not just financial; the close relationship can have value in the longer run too, providing insight into the developing technology, business implications and opportunities.
Reuters’ latest readers survey reveals a record new low in trust in news – 37% in the UK. Legacy media brand trust remains strong, but now older audiences demographics are following the young into getting their news from third party platforms.
What about AI for news? Those getting their news from AI are a small percentage globally, from the ‘AI superusers’, and as publishers already know, clickthrough rates are lower than for social.
Could AI for news discovery grow? People tend to ask AI specific questions. I think it’s a slightly odd thing to ask an AI chat for ‘the news’, and you’d have to check that what it’s telling you is true, which means going to a trusted news site anyway. But, perhaps, many people don’t care much about reliability – they don’t trust it, but that’s not that important to them. And I can imagine AI providing news by default on a search home page – like MS News and other browsers did 25 years ago.
And before that, I can imagine, and publishers might like, human subscribers using their own AI agents to send bots to their sites, perhaps a number of publishers’ sites, to pick content from one site or to go further and build their own digital newspaper or magazine. That means publishers have to let those ‘good bots’ in while blocking the ‘bad bots’.
That use of agentic AI could be extended to company’s agents, especially in the B2B market. The big AI platforms could be licensing at one end of the market. The numbers are low right now, but could that grow under pressure from industry cooperation like SPUR, which includes the giants of broadcasting, national news and for magazines, the PPA? Let’s hope so. But few trust the tech giants to do the right thing. At the other end of the market, the individual readers’ personal agent could be made ‘sub-ready’. But in the middle of this future agentic economy is a potentially interesting middle-market of companies looking to buy content for their businesses. They can only really buy in the scraping economy at the moment, but there could be opportunities for publishers there.
Another way through for publishers, the return of an older strategy, is to build the personal relationships in the models around content. That means knowing your reader – even in the old-fashioned news desk way of sticking a caricature drawing of ‘our reader’ on the wall, and the return of building communities. And emphasise the ‘proud to be human angle’, satisfying any anti-tech backlash looking for the ‘authentically human’. From these, new forms of subs and memberships will emerge. We hope.
AI search and agentic AI are the future yet the discussion sounds eerily familiar.
The very worst-case scenario remains a terminator style destruction of the human race. I wrote about that in this column in June 2023 but it’s not gone away. If AI company insiders are to be believed and it’s not just marketing hype, it’s a serious possibility – 10% to be precise. But there’ll be no one around to report it. Except the agentic AI reporters perhaps. But no humans to read it. Except the AI agentic readers.
On the bright side, there’s a 9/10 chance we’ll survive – both as a media industry and as a species.
This article was first published in InPublishing magazine. If you would like to be added to the free mailing list to receive the magazine, please register here.
