AI Search & Some Implications for Publishers and the Public Sphere
It’s only Thursday and it has already been a big week for AI announcements. I haven’t quite made up my mind about it, but here are some initial thoughts about the implications for tech companies, publishers, and the public.
Both Google and OpenAI seem to double down on “agentic” AI and multimodality. Seamless voice interaction and search is not new, of course, but if the demoed near-natural conversations without a significant lag prove to be reliable, interesting times could be ahead.
I am still not super convinced that voice interaction is the big flex tech companies think it is. People seem to generally like to consume written information (higher information density, easier to store/re-use, better privacy protection), but it might make a dent in traditional search and the way we interact with information.
More interesting to me is how Google is doubling down on bringing AI to search. It’s something many expected for months, and now it’s happening. Davey Alba and Julia Love have a good summary over at Bloomberg here or you can get it directly from Google’s blog.
For now, we don’t know for which searches SGE will kick in. Industry analyst Bart Horalewicz has some interesting data on this, suggesting that so far a majority of searches do not trigger SGE. But who knows if it will stay that way.
The Associated Press quotes an industry expert as saying that about 25% of traffic to publishers could be negatively affected by the de-emphasis on website links, with knock-on effects on ad revenue and subscriber numbers. All this comes against the backdrop of already declining referral traffic from platforms. As Press Gazette documents (h/t to Peter Erdelyi’s newsletter), it has gone down for both Google and Facebook. See their reports on Facebook’s referral traffic and the Google Core Update.
With respect to their AI search SGE, Google claims that “we see that the links included in AI Overviews get more clicks than if the page had appeared as a traditional web listing for that query.” We have to take that at face value for now. Whether that will be borne out by the evidence is another question. The claim, of course, is that they will keep on sending traffic to publishers, but who knows how much. And they are not saying anywhere that traffic won’t decline (as far as I can tell).
The question for publishers is, of course, how they can ensure that people will still 1) find and 2) consume their content (and for which reasons and needs) in an environment where many people already find many of their information needs covered elsewhere. Reed Albergotti writing for Semafor put it more bluntly: “Publishers […] need to make tough decisions. How can they build businesses that they alone control?” How can they, indeed?
No matter how much publishers and critics wail, that’s of little help if users come to like this technology (and here being the de-facto monopolist over search helps a lot. Standard-setting power for the win…). As I put it to Melissa Heikkilä and MIT Tech Review: “This will help speed up the search process and get users more specific answers to more complex, niche questions […] I think that’s where Search has always struggled.” And answers don’t have to be perfect, just good enough in many cases. See Kevin Roose for NYT: “It may also be a popular feature with users — I’ve been testing AI overviews […] and have generally found it to be useful and accurate.” (Full article here).
And the question about the training data and copyright remains. Where does the training data come from? Who should be compensated and how? We never hear much about this in these announcements — instead, we get lots of safety and brave new world talk. This is also a task for regulators who have basically let this happen because of “innovation.” At least the House of Lords Digital Committee didn’t mince words in its latest report:
The big picture
It’s another step-change in the competition for audiences and attention. I think we will see a further tethering of people more closely to information and to these companies as providers of information (including news), with platforms taking over even more of the informational needs that people used to get through, e.g., news or other content on the web.
By giving users summaries instead of directing them to the original sources, it will likely take away the attention and money-making opportunities for those who create and invest in producing original information (such as publishers)… which wouldn’t be great.
Not only could we see a “flattening” and homogenization in terms of what users get to consume (one “average” answer instead of a diversity of results) by virtue of how these systems work; if the input becomes worse, it reinforces the issue. Political Scientist Andreas Jungherr has described this nicely over on Twitter/X: it poses a threat to the idiosyncrasies of signals in the public arena.
As I put it in my recent Tow Center report, all this ultimately provides a pathway for tech companies to further cement their control over information, and potentially enabling them to take over tasks that were once central to, e.g., journalism. Control here to me means: the ability to (1) shape the structural conditions of the public arena (Who gets to see what? Who gets heard? Who gets to organize information?) in (2) the service of economic aims (who gets to benefit?) and connected to (3) political aims (which are often about shaping regulation to entrench and extend one’s control and position).
And the race to consolidate and extend this control is heating up.
(Last updated 16 May 2024)
Felix M. Simon is a Research Fellow in AI and Digital News at the Reuters Institute for the Study of Journalism at the University of Oxford.