Google presented on Tuesday its latest innovations in generative artificial intelligence (AI), which will potentially transform the daily lives of its users, from online search to many everyday tasks, thanks to ever more omniscient assistants.
• Read also: OpenAI gives new superpowers to ChatGPT
• Read also: ChatGPT: 7% of Quebec students have already used it to do an assignment for them
• Read also: OpenAI competes with Google?
“We are still at the very beginning of the transition to the era of AI,” immediately recalled Sundar Pichai, the boss of Google, on the stage of the open-air amphitheater of the company in Mountain View, California.
The American technology giant held its annual conference for developers on Tuesday, under the banner of generative AI (production of content on a simple request in everyday language).
According to Sundar Pichai, “the most exciting transformation is obviously generative online search on Google.”
Google has already been testing its new approach for a year: at the top of the results page, the Internet user receives a written answer to their question, generated by Gemini, Google’s AI model. He can then click on suggestions for additional questions, or, further down, on traditional links to websites.
Conclusive tests, according to the manager: “users do more research and are more satisfied,” he assured.
The new formula – the most significant transformation of the search engine since its creation – will be deployed in the United States this week, then in other countries, to reach more than a billion people by the end of the year.
“In the age of Gemini (…) Google does the work for you,” promised Liz Reid, head of Google Search.
She showed how new generative AI tools are making users’ lives easier, whether they’re looking for a yoga studio or planning an entire trip.
Research under threat
Google dominates online search to the point that its name is synonymous with the action.
In early 2023, thanks to its massive investments in OpenAI (ChatGPT), Microsoft added generative AI to Bing, its search engine. In vain: Google remained the reference.
But the current technological revolution could still threaten it. All of Silicon Valley is competing for new AI tools and assistants, which make it possible to bypass the world’s number one digital advertising company.
On Facebook, Instagram and WhatsApp for example, users can ask questions to Meta AI, which now has access to the internet.
The strategic analysis firm Gartner predicts that by 2026, the volume of queries to traditional search engines will fall by 25%, as chatbots and AI assistants like ChatGPT and others eat away at share. Steps.
The battle therefore shifts to these digital assistants, which seem to gain new superpowers every week.
Particularly thanks to advances in generative AI models.
That of Google, Gemini 1.5 Pro, will be able to take into account more context information provided by the user (hundreds of pages of text, longer videos, etc.) and gain in multimodality (the model “understands” as well text, sound and images, and can respond in writing, by voice or by generating images).
“In-Depth Conversations”
Beyond technical capabilities, Sundar Pichai presented his vision for the future: AI agents will be “intelligent systems capable of reasoning, planning and retaining information. They anticipate the steps and know how to work with software to accomplish things on your behalf, under your supervision.
The company, which made nearly $74 billion in profits last year, is investing across the board to bring this vision to life.
On Tuesday, she presented Gemini Live, which will allow people to have “in-depth conversations with Gemini”, orally, via the mobile application.
Later this year, the assistant is expected to gain skills thanks to advances in Project Astra, a prototype AI agent.
Google’s research lab, DeepMind, released a video demonstration of Astra, which was enthusiastically received by the public.
We see a user pointing the camera of his smartphone – or glasses with an integrated camera – on his environment, and interrogating the model, which correctly identifies the place where he is located, solves a computer problem based on a diagram and remembers where he put an object.
On Monday, OpenAI made a similar presentation, where ChatGPT interacts with an engineer in a way so natural that the machine seems human.
But the fluidity of conversation does not yet make chatbots omniscient, proactive and personalized AI agents.
Google DeepMind introduced other new models, Gemini 1.5 Flash (faster and cheaper), Imagen 3 (image generation) and Veo (video generation, a growing sector of generative AI).
