The counter-offensive will not have dragged on.
While ChatGPT has been in the headlines for several weeks, Google unveils its own conversational artificial intelligence, named Bard, on Monday evening.
As with the tool published by the American company OpenAI, Bard himself composes a text to answer questions asked by the user.
But the two services have weight differences.
Unlike ChatGPT, Bard is connected to the Internet "
to provide up-to-date, high-quality answers
," said Sundar Pichai, CEO of Google, in a blog post.
Another distinguishing feature: Bard is not accessible to the general public, at least initially.
“
We are opening Bard to trusted testers, before making it more widely available to the public in the coming weeks
,” continues the leader.
Read alsoFrom Dall-E to ChatGPT, how OpenAI has democratized a new part of artificial intelligence
This two-phase launch will “
ensure that Bard's responses reach a high standard in terms of quality, security and grounding in real-world information.
This testing phase allows us to continue learning and improving the quality and speed of Bard.
This is Google's response to one of the main criticisms of ChatGPT: the OpenAI service can assert erroneous facts with great aplomb.
Unlike the young Californian company, Google cannot afford to market an imperfect product that would harm its credibility.
Synthetic answers in the search engine
Bard relies on the LaMDA natural language model, which Google has been working on for several years.
It is in the name of this reputational risk that the Mountain View group had restricted access to LaMDA to its engineers only, the time to perfect this artificial intelligence and to ensure that it does not present any risks.
One of these employees, Blake Lemoine, made headlines this summer by declaring that LaMDA had a conscience... He was fired.
But the global buzz around ChatGPT, coupled with billions of dollars invested by Microsoft to integrate OpenAI technologies into its own products, has led Google to accelerate its plans.
Other innovations related to LaMDA will thus soon appear on the Google search engine, announces Sundar Pichai.
Read alsoBehind ChatGPT, the shadow of the giant Microsoft
"
Internet users usually turn to Google for quick factual answers, like 'how many keys does a piano have?'
However, more and more yearn for more in-depth information and understanding, for example: “Which is easier to learn between piano and guitar, and how much time does each take to practice?”
"
AI can be helpful in times like these, synthesizing information for questions that
don't necessarily have a single right answer.
"
According to the example provided by Google, a text generated by LaMDA will provide a synthetic but nuanced response to the Internet user and redirect him to several sources of information on the subject.
A similar feature could soon arrive on Bing, Microsoft's search engine.
The next version of Google will display synthetic answers to certain complex Google queries
Combining innovation and responsibility
Last announcement, Google will allow companies and independent developers to access certain features of LaMDA in March in order to create their own solutions.
“
Over time, we intend to build a suite of tools and APIs that will make it easier to build more innovative applications with AI
,” Sundar Pichai points out.
“
This is just the beginning.
New announcements are to come in the weeks and months to come
, ”continues the leader.
According to the American press, the popularity of ChatGPT has prompted Google to accelerate the bringing to market of concrete tools powered by generative artificial intelligence.
The Californian group seeks to reconcile two imperatives: “
to be audacious in terms of innovation and responsible in our approach.
»