Photo-Illustration: Intelligencer; Photo: Getty Images

Early this year, Google teased a fundamental change to its core product, the search engine through which much of the world accesses the web. Soon, the company said, Google would start using AI to “distill complex information and multiple perspectives into easy-to-digest formats.” By May, the company had a real product to share.

For Google, it was an obvious and incremental feature update combining two of the company’s products: a text generator plugged into a search engine, basically. Searchers ask a question, and Google tries to answer it with short, article-style “snapshots.”

For publishers, however — of news, how-to content, reviews, recommendations, reference material, and a range of other content one might describe as existing to “distill complex information and multiple perspectives into easy-to-digest formats” — it looked like nothing less than an existential crisis. Google was getting into content, automating the work of its partners, and dramatically altering the terms of its informal deal with publishers that has sustained digital media for years: You make content; we send traffic; everyone sells ads. If this wasn’t a threat to journalism directly, it was certainly a threat to the journalism business. Google, it seemed, was eager to cut the publishers out.

It’s early, still, and AI search won’t threaten much of anything if it fundamentally doesn’t work, or if users don’t like it, which we’ll know soon enough. But it doesn’t have to be perfect, or even great, to dramatically alter the online economy. A stickier question is whether Google, possessed of a new capability to inflict massive harm on digital publishers and the web in general — and meanwhile battling very different firms for AI dominance — will decide, in the coming months, that it is in its own business interest to do so.

In its current form, Google’s Search Generative Experience will answer a question about the debt ceiling with a lengthy attempt to summarize the news.

Up top, searchers get a 272-word summary of the news with a bit of background. Its citations, which are hidden behind a small button in the upper-right portion of the screen, include a consulting firm, a think tank, and a slew of news organizations, including the New York Times, The Wall Street Journal, and NBC. Conventional search results are well beyond the bottom of the screen; on this issue, the information was accurate, though it’s still pretty easy to get tripped up.

Media executives are sounding the alarm. “Our content is being harvested and scraped and otherwise ingested to train AI engines,” said News Corp. CEO Robert Thomson at the INMA World Congress of News Media last week. “These are super-snippets containing all the effort and insight of great journalism but designed so the reader will never visit a journalism website, thus fatally undermining that journalism.” He added, “Content mining is an extractive industry.” Brian Morrissey, the former editor of the media trade publication Digiday, outlined publishing’s Google predicament at The Rebooting, predicting the decline of the web page in general:

As Google eliminated all credible competition, search became a mostly reliable distribution channel. The bargain was always for publishers to play by Google’s rules, then make money from ads that very often ran through Google’s ad stack and let them wet their beak. It was a roundabout way of paying tribute to the king. Nobody likes taxes, but if someone controls the distribution, you pay up …

That’s breaking. Google’s demo of its new AI-fueled search engine heralds a new phase of search that will throw the page’s central role in publishing strategies into question.

“From Google’s demos, what’s clear is less traffic will go to publishers,” he said. Less traffic means less of everything that keeps modern media companies afloat: advertising revenue, subscription conversions, e-commerce revenue.

“At the risk of overstating the potential consequences,” wrote Matt Novak at Forbes, Google’s search overhaul “will be like dropping a nuclear bomb on an online publishing industry that’s already struggling to survive.”

Google stressed that this was an experimental feature and that, for now, it would be limited to testers who opted in. Certain categories of queries would not trigger the snapshots, the company said — sensitive medical questions, for example — and each answer can be checked, sort of, by clicking a button that reveals linked citations for each sentence. Classic results would still be present, though less visible.

Still, the change would represent a fundamental shift in what Google does, how users interact with it, and how it interacts with the web around it. For billions of people, Google is the default interface for the rest of the online world. It’s the portal through which all other sites are accessed. It’s the box — on your phone or your computer or your tablet — with which you interact so often you take it for granted. It’s a de facto governing authority for the parts of the internet that aren’t hidden away inside social platforms and apps and has unparalleled sway over what gets seen online and by how many people. If implemented at all, by virtue of Google’s size, it would have a significant effect on traffic for pretty much any digital publisher.

This is a facet of the larger AI story — which is to say it’s about automation. But it’s also a story of a large platform deciding to compete more aggressively in the marketplace it controls. With snapshots, Google is pushing into some of the most lucrative parts of the content business over which it already exerts enormous influence. That the sorts of content it seems to be automating first are explainers, guides, and product rankings is no coincidence — these are styles of content that publishers currently produce with Google traffic in mind. If Google hired tens of thousands of contractors to produce “snapshots” and product recommendations for popular searches, it would be easy enough to conceptualize and very bad news for a number of Google-dependent online industries; that it’s doing so with “generative AI” suggests that what was holding it back from attempting to replicate or replace some of the most trafficked sites on the web wasn’t some lofty notion of how Google should function as a market or an ecosystem, some sense of stewardship over “the web” as a concept, but cost.

A lot of dark predictions about AI are counterintuitively sort of naïve, imagining the technology as a distinct and novel entity with its own motives or as a phenomenon that will be evenly experienced across the economy. Google, here, teases a more familiar story, utterly devoid of novelty: Large firm seeks efficiencies and uses machines to achieve them.

The doomsayers have a point, in other words: If Google commits to summarizing more and more of the content it used to serve, the companies that make it are in for an even worse time than they’re already having. The vast majority of publishers are individually insignificant to Google and have no collective power to speak of. With apologies to Mr. Thomson, News Corp. properties, with their search-engine-optimization teams and content strategies, are already scrounging for traffic from the margins of Google’s user experience. As any SEO professional will tell you, it wouldn’t take something so dramatic as an “AI-search makeover” to lose a significant chunk of your inbound readership from Google. Small mysterious updates to its search algorithms have pitted publishers against the company’s machine-learning systems for years.

In publishing, however, there is also a tendency to overestimate the forecasting abilities, and general competence, of larger and more successful technology companies. Google, one of the largest tech companies in the world, has a lot to gain and lose by altering search, which generated $162 billion of Google’s $224 billion in advertising revenue in 2022. It has skin in the game. Will Google users be happy with a machine-improvised Wikipedia article at the top of their search results? Will it change their relationship to the sponsored links at the heart of Google’s business? Will they take product recommendations seriously from a Google bot? Will Google’s AI testing phase result in doubling down on content automation or quietly rolling it back? Will that be because users don’t care for it, or because they do, but it’s in a way that threatens Google’s business? Their predicament is the AI dilemma in not-so-miniature: a confrontation with the essential weirdness of generating synthetic information.

Replacing outbound links to the web with machine-synthesized summaries of the web is both an obvious use case for generative AI and a direct threat to the economy in which a range of content — including journalism — is currently produced. But its success depends on a few assumptions: that the summaries are good or, far more important, that people think they’re good and trust them; that, in the long term, there remains sufficient scrape-able content to summarize; that the web ecosystem Google will be exploiting won’t be itself overrun with AI-generated content, leading to a death spiral of content credibility and relevance; that stepping deeper into the content business makes any sense for Google, the leadership of which might be acting out of fear of missing out on the next big thing, at the company’s peril. Some of these issues are less speculative than others. For decades now, the entire web has been optimizing itself for Google, modifying and producing content with search traffic in mind; Google, which was built around the idea of surfacing and organizing the world’s information, has instead created the mother of all spam problems, which it struggles daily to solve.

But from the user perspective, Google as an AI-powered answer engine is also uncharacteristically aligned: It casts present-day Google Search as something broken that needs to be fixed — which, well, maybe it is. Rather than contending with a cluttered interface and a gauntlet of advertising to get to a credible link, the company has teased something clean, clear, and refocused on results. The company’s AI-search demos have doubled as scathing critiques of the mess that search has become and of a business model that depends on interruption, diversion, and extra engagement. Maybe this pristine alternative vision is indeed what we end up with, in which case the web as we knew it is shoved off the page, a decades-old online civilization of websites reduced to training data for slick chatbots.

Or maybe, after a brief detour, Google’s true identity as an advertising business reassumes control and once again draws it, and its users, back into the lucrative mess, where they will continue to tap and click their way through interfaces that are designed as much to monetize them as to assist them in anything resembling a “search.” For Google, it might be better to have a web to exploit than to have no web at all.

Adblock test (Why?)

Source link