From our correspondent in California,

When OpenAI unveiled the power of ChatGPT, and Microsoft felt the tide turn by signing a major partnership to catapult its Bing search engine into a new era, Google declared a "code red" internally. On Wednesday, at its Google I/O developer conference in Mountain View, we discovered the first concrete results of this mobilization. Launch of its Bard chatbot in 180 countries, emails written by the machine, synthetic images... Google is betting on artificial intelligence, while promising to use this technology in a "responsible" way. Especially for research, with which the giant is particularly conservative.

The Bard chatbot accessible in 180 countries

The waiting list is over. Two months after its hasty response to ChatGPT, Google is launching its Bard smart chatbot in 180 countries. For now, it's only in English, Japanese and Korean, but Google promises support for 40 languages, including French, for "soon".

Bard builds on the power of PaLM 2, Google's latest LLM (large language model), and can allow people without technical knowledge to create computer programs by describing the desired result, such as "Python code chess piece moves". Google has announced partnerships, including with Adobe, to generate images with Bard, for example to create a birthday invitation for a little girl who loves unicorns.

Gmail able to write emails on its own

Gmail was already suggesting short answers. With "Help me write", the software becomes a real secretary. Need to write an email to request a refund? He not only deals with the text, but goes looking for all the factual elements in our correspondence. If the result is too polite, he may be asked to change the tone.

Clearly identified synthetic images

We are starting to see synthetic images that are larger than life, which poses big problems for misinformation. In its accountability principles, Google has promised to clearly identify the images generated by its AI. Images, but also videos, such as its universal translator, which is able to perform dubbing in near real time, with mouth movements modified to match the new syllables.

A dose of AI in search results

It is finally on the heart of its business – and its business – that Google is the most timid. AI is landing in its search engine, but slowly, in beta, and only in the United States, by registering in Search Labs. This "Search generative experience" (another wooden name from Google) will be launched in a few weeks.

Access to this content has been blocked in order to respect your choice of consent

By clicking on "I ACCEPT", you accept the deposit of cookies by external services and will thus have access to the content of our partners

I AGREE

On the program, a response digested by the machine, after passing the Web to the mill, which is displayed above the classic blue search results. Google has been careful to avoid giving AI a personality, which remains very factual and hopefully will avoid hallucinating or inventing quotes.

It's experiential for now, but the Mountain View company has already integrated ads, such as sponsored results for "a bike adapted to a hilly 8 km journey". Google's biggest challenge is not to prevent a machine uprising but to prevent a competitor, or AI, from killing its goose that lays the golden eggs.

  • Tech
  • Google
  • ChatGPT
  • OpenAI
  • Microsoft
  • California