The arrival of AI on the Google engine raises concerns


Liz Reid, vice president of Google, in Mountain View, California, Tuesday, May 14, 2024. JEFF CHIU / AP

“Let Google do Google searches for you. » It is with this slogan in the form of a pirouette that Liz Reid, vice-president of the company, summarized an important change announced Tuesday May 14: the leader in search engines is integrating an assistant which will generate, thanks to intelligence artificial (AI), responses to Internet users’ queries.

The interface is reminiscent of that of existing conversational robots, such as ChatGPT or Gemini, from Google, but it is incorporated into the American flagship service, search. This development is being rolled out first in the United States over the coming weeks, then in other countries by the end of 2024, with the aim of reaching “1 billion people”. This turning point also raises questions and concerns.

Read also | Google will slip its Gemini AI into Android and its search engine

Add to your selections

Concretely, for complex queries or questions – like “What constellations are easy to observe in the sky? » –, the engine will respond with a few paragraphs written by the AI ​​assistant, along with a few related links to sites. The list of links usually presented by Google will be displayed further down.

Read also: Article reserved for our subscribers Google suggests checking what its Bard robot writes to limit its errors

Add to your selections

Risk of “response engine”

More advanced or thematic functions will be added. The assistant therefore offers to plan meal recipes or trips. And, ” Soon “, the AI ​​will respond to “queries requiring multiple searches”, specifies the company, such as selecting yoga salons within a ten-minute walk. Finally, the Internet user will be able to question the assistant orally or integrate a photo and, soon, a video into their request.

Google sees this as a “platform change” comparable to the transition from fixed to mobile Internet. For the company, it asks questions about the reliability of the answers (the assistants all make factual errors), but also about the place of advertisements on its engine (which earned it 175 billion dollars, or around 161 billion euros, in 2023, more than half of its activity), even on its economic model (each request to an AI costs a few fractions of cents, and the company did not exclude reserving certain advanced functions for a paid subscription).

Read also: Article reserved for our subscribers The arrival of ChatGPT or Bard on search engines worries website publishers

Add to your selections

Above all, since its announcement a year ago, this change has generated fears among website managers. “It’s a change of era”, estimated in Wired Jim Yu, founder of BrightEdge, who advises publishers on how to optimize their position on Google. “The risk is that the search engine becomes a response engine. And that AI-generated responses take the place of website links on the engine. This would be economically catastrophic for us”, judge Pierre Petillault, general director of the General Information Press Alliance, also asking the question of “transparency” conditions for generating responses. With AI, Google would become more of a “closed garden”, with research generating “zero clicks”. Which would justify all the more, in the eyes of the media, to negotiate remuneration for the use of their content by AI.

You have 21.91% of this article left to read. The rest is reserved for subscribers.



PREV Tour of Italy – 11th stage – Another sprint victory for Jonathan Milan, Tadej Pogacar still in pink
NEXT Slovak Prime Minister Robert Fico shot and wounded; ” His days are endangered “