"We've been primarily an artificial intelligence company for seven years and we're at a turning point," Sundar Pichai, the boss of the Californian group, told thousands of people gathered in his amphitheater in Mountain View.
"For some time now, we have been making our products radically more useful through generative AI, following a bold and responsible approach," he added.
Google, lagging behind Microsoft on the deployment of generative AI, is to present Wednesday its latest technological advances to developers and journalists © Pau BARRENA / AFP / Archives
The November release of the ChatGPT interface — designed by California-based startup OpenAI, mostly funded by Microsoft — launched a frantic race for generative AI, between exuberant enthusiasm and apocalyptic worries.
Google responded with its own interface, Bard, which opened to the public at the end of March. On Wednesday, the company announced that the chatbot is now available in English in 180 countries around the world.
The chatbot will soon be able to converse in 40 languages and must become multimedia, i.e. be able to integrate images into users' questions and answers.
The world leader in digital advertising also showed how online search will gradually change, with written answers to users' questions above traditional links, and the possibility of exchanging with the interface, asking for details, for example.
The new Google is due to arrive in the coming weeks, the company has opened a waiting list on which you must register to use it in a few weeks.
The Californian group is also developing extensions to Bard, so that users can interact with the robot directly from the Maps mapping application, the Gmail mailbox or the online word processor Docs.
Sundar Pichai, the boss of the Californian group Google on Wednesday, May 10, 2023, in front of thousands of people gathered in his amphitheater in Mountain View © Josh Edelson / AFP
Customers of its cloud business are not forgotten, with tools for companies that want to design their own generative AI-based tools (search, chatbots, etc.) for their applications, with their own data.
Microsoft recently made similar announcements. The IT firm had already integrated ChatGPT into its Bing search engine and completely opened it to the general public last week, relaunching this portal so far negligible compared to Google.
The two competitors compete with ads with a stated goal: that their AI-inflated platforms become the privileged personal assistants of users.
Google on Wednesday presented PaLM 2, the new, more advanced version of its language model, these algorithms trained on mountains of data that make it possible to create conversational robots like ChatGPT or Bard.
"This has allowed a lot of improvements in Bard in the last two months, in terms of mathematics, logic, ability to reason or code and correct code," said product director Jack Krawczyk at a press conference.
"We are at a fascinating moment where the gap between human imagination and the capabilities of technology is rapidly closing," he added.
But the prowess of Silicon Valley is worrying, especially since OpenAI launched in March GPT-4, a "great multimedia model (...) as successful as humans in many professional and academic contexts."
The boss of the start-up, Sam Altman, explained working towards the so-called "general" artificial intelligence, that is to say programs with human cognitive abilities.
Since then, many experts have voiced fears, from misinformation to job replacement, going so far as to call for a six-month pause in research.
How ChatGPT is trained to answer © us Valentin RAKOVSKY / AFP/Archives
Geoffrey Hinton, considered one of the founding fathers of AI, said on May 3 that the "existential threat" that AI posed to humanity was "serious and close", during a round table organized by the Massachusetts Institute of Technology (MIT).
The computer scientist has just left Google, where he had worked for ten years, "to be able to talk about the dangers of AI" freely, he explained.
© 2023 AFP