The company triggered the “red code”, asked for help from Sergey Brin and Larry Page, and plans to launch around twenty projects around generative artificial intelligence, to deal with the threat of the chatbot.
They are back. Sergey Brin and Larry Page, the two founders of Google, have been called to the rescue, reports the New York Times. The creators of the firm, who left the management in 2019, participated in several meetings with senior executives. The reason for this operation worthy of an American action movie? The release of a certain ChatGPT at the end of November 2022 by the company OpenAI. This artificial intelligence-enhanced chatbot is causing a wave of panic among the Web juggernaut.
“Ask ChatGPT” is the new “Ask Google”?
It must be said that the adoption of this text-generating AI was very rapid. A week after its launch, more than a million users had already tested it. Since then, Internet users have multiplied the experiments. ChatGPT, capable of laying out a text on any subject in a few seconds, is used to write essays, correct lines of code, write press articles… AI is far from being faultless, but the most convinced see it as a new way to search for information online. What give a few cold sweats to Google, in a situation of quasi-monopoly for two decades.
Sundar Pichai, CEO of Google, even triggered the “red code”. All ongoing projects have been suspended in order to focus on generative artificial intelligence. According to an internal presentation that the New York Times obtained, the firm plans to present 20 new projects and prototypes related to AI this year, as well as a new version of its search engine integrating chatbot functionalities. During the meetings in which they participated, Sergey Brin and Larry Page would have submitted some ideas and validated some of these prototypes in progress.
Among these projects are Generation Studio, which creates and edits images; an application called Maya which allows you to visualize shoes in three dimensions; or a tool capable of summarizing videos by generating a new one, depending on the slides seen by the New York Times.
Small problem: with ChatGPT, no targeted advertising
Google panics, but the company has not remained idly by in the face of recent advances in artificial intelligence. The firm has developed its own in-house model similar to ChatGPT (presented before the OpenAI model was put online). He is baptized LamDA, to Language Model for Dialogue Applications. Small problem if Google replaces its classic search engine with a tool of this type: what about targeted advertising – which represents more than 80% of its income?
Another problem: for the moment the text-generating AIs are far from reliable. They rely on huge textual databases (including billions of web pages) and mimic text written by a human. Sometimes the background is true, sometimes partially or completely false. Slightly problematic for a tool supposed to deliver information. Not to mention the toxic remarks and other racist or sexist biases that can be found in their productions. Admittedly, Google can bring up links disclosing false information and other toxic words, but the question-answer side of ChatGPT gives it even more problematic insurance. On Google you can still make your own market, with a so-called expert AI, it’s more difficult to question the answer given.
Many researchers specializing in these models also call for caution, and do not advise using ChatGPT or its equivalents to obtain information. As a new company, OpenAI can afford to launch an artificial intelligence that is still experimental and perfectable. For a company as established as Google, it’s much more complicated. Meta had the painful experience of this when in November it launched Galactica, a chatbot producing scientific articles from a short text. Mark Zuckerberg’s firm had to remove AI from the Web three days later, after numerous critics pointing out its errors and absurdities.
Mass layoffs at Google
Despite these obvious limitations, the race to see who has the biggest generative AI seems well and truly on. And Google has no choice but to mobilize on the subject. Besides OpenAI, other companies are trying to compete with the search engine with text-generating AI. This is particularly the case of You.com, a new search engine, equipped with a YouChat functionality very similar to what ChatGPT can do.
Will 2023 sound the death knell for Google’s omnipotence? We’ll see. In any case, it is certainly one of the most important challenges that the company has had to overcome. “This is a moment of significant vulnerability for Google,” said D. Sivakumar, a former Google research director in New York Times. ChatGPT gave netizens a glimpse of what an “engaging new search experience” could be .
For the moment, the company, like its Silicon Valley counterparts, is managing the return to reality after the period of hypergrowth linked to the confinements. Google announced on January 20 the loss of 12,000 jobs.