Can Google provide AI answers without harming the web?

For the past year and a half since ChatGPT was released, a frightening question has hovered over the heads of major online publishers: what would happen if Google decided to overhaul its main search engine to give greater prominence to generative artificial intelligence – and in would it interrupt our business in the meantime? ?

The question speaks to one of the most fragile dependencies in today's online media ecosystem.

Most large publishers, including the New York Times, get a significant share of traffic from people who go to Google, search for something, and click on articles about it. That traffic, in turn, allows publishers to sell ads and subscriptions, which pay for the next wave of articles, which Google can then show to people searching for the next thing.

The whole symbiotic cycle worked well, more or less, for a decade or two. And even when Google announced its first generative AI chatbot, Bard, last year, some online media executives consoled themselves with the thought that Google would never put such patchy and unproven technology into its search engine , or risk ruining his profitable research. advertising business, which generated $175 billion in revenue last year.

But change is coming.

At its annual developer conference on Tuesday, Google announced that this week it will begin showing AI-generated responses — which it calls “AI overviews” — to hundreds of millions of users in the United States. More than a billion users will receive them by the end of the year, the company said.

The answers, powered by Google's Gemini AI technology, will appear at the top of the search results page when users search for things like “vegetarian meal prep options” or “day trips to Miami.” They'll provide users with concise summaries of what they're looking for, along with suggested follow-up questions and a list of links they can click to learn more. (Users will still receive traditional search results, but they will have to scroll further down the page to see them.)

The addition of these answers is the biggest change Google has made to its main search results page in recent years, and stems from the company's fixation on putting generative AI into as many products as possible. It may also be a feature users appreciate: I've been testing AI overviews through Google's Search Labs program for months and have generally found them useful and accurate.

But publishers are right to be scared. If the AI-powered response engine does its job well enough, users won't need to click on any links. Whatever they're looking for will be right there at the top of the search results. And the great pact on which Google's relationship with the open web is based – you give us articles, we give you traffic – could fall apart.

Google executives put a positive spin on the announcement on Tuesday, saying the new AI-powered overviews would improve the user experience by “eliminating the work of searching.”

But this legwork pays off in a lot of journalism and many other types of online media (fashion blogs, laptop reviews, restaurant listings) without which the Internet would be much less useful. If Google's AI overviews starve these websites of traffic, what will happen to them? And if large portions of the web vanished entirely, what would be left for artificial intelligence to summarize?

Google had clearly anticipated these fears and its executives had already prepared responses.

In a briefing this week ahead of Google's developer conference, they said the company's tests had found that users who were shown overviews of the AI ​​tended to conduct more searches and visit a more diverse set of websites They also said that links displayed in AI overviews got more clicks than links displayed in traditional search results pages.

Liz Reid, Google's vice president of search, said in a blog post Tuesday that the company “will continue to focus on sending valuable traffic to publishers and creators.”

But analyze these responses carefully and you'll see that Google isn't saying publishers' overall search traffic won't decline. That's because Google can't really predict what will happen once it starts showing AI-generated overviews in billions of search results a day, and how user behavior might change as a result.

Earlier this year I wrote about Perplexity, an AI-powered “answer engine” that shows users a concise summary of a topic they're researching instead of giving them a list of websites to visit. I believed the experience was clearly better than a traditional search engine for some types of searches and usually provided me with more useful information faster.

But I was also nervous, because during my tests of Perplexity, I pretty much stopped clicking on any links. In a world where AI can browse the internet for me and paraphrase what it sees, I found that I simply didn't need it. And I worried about what would happen if Perplexity users were all like me and got into the habit of relying on AI-generated summaries rather than original sources.

I have the same concerns about Google's new AI overviews, but on a very different scale.

The perplexity is minimal: only 10 million monthly users in February. Google, by contrast, has billions of users and accounts for more than 90% of the global search market. If they make a change to their search engine that reduces outbound traffic by even a few percentage points, every publisher will notice.

It's unclear how big the effects of Google's AI overviews will ultimately be. One analytics firm, Gartner, has predicted that traffic to the web from search engines could decline by 25% by 2026. And many publishers are bracing for double-digit drops in traffic this year.

Perhaps these fears are exaggerated and the publishers were worried for nothing. But after Tuesday's announcement, Google made it clear that they're about to find out anyway.

Leave a Reply

Your email address will not be published. Required fields are marked *