Bard is Google’s chatbot. Made official in February, it is supposed to respond to OpenAI’s ChatGPT, which Microsoft wants to integrate into Bing and Edge.
What is Google Bard?
Google Bard is a chatbot, that is, a conversational agent, similar to OpenAI’s ChatGPT. As the name suggests, it is designed by Google. Its role is to generate answers to questions or instructions given to it by the user — these commands are called “prompts”, in other words orders given to it.
Like other chatbots, Bard is called “artificial intelligence”. It is a formulation accepted in everyday language, but which is not technically exact. Of course, Bard mobilizes algorithms and processes from the field of research on artificial intelligence. However, it is not a real AI, like Skynet, Jarvis, HAL 9000 or C-3PO.
Bard specializes in generating content in the form of text (there are other tools that are tailored to produce something else, like music or images). It is based on the LaMDA language model and, thanks to its access to the Internet, it can exploit the most recent information to respond to requests from the public.
In Google’s plans, Bard is to be integrated into its search engine soon, perhaps in place of the “I’m lucky” button on the homepage. Today, the chatbot is still in the running-in phase, in order to ensure the proper behavior of the tool. It would be terrible to offer a chatbot chaining errors or absurd answers.
Why did you choose Bard to name this chatbot?
Google did not opt for a complex name, such as ChatGPT, which is a portmanteau combining “chat” to evoke a discussion and GPT, an acronym describing the technique used by OpenAI to direct its chatbot. The search giant has rather summoned the collective imagination, drawing inspiration from… bards, these poets and storytellers from Gaelic culture.
THE New York Times reported on Feb. 6 that Bard was held back because Google imagined his chatbot as a “ storyteller “. The name is elegant, but perhaps slightly out of step with what one expects from a tool like Bard: one does not want to read stories or fables so much, but to have access to reliable information.
When was Bard made official?
Google formalized the existence of Bard in the week of February 6 and, in the process, the search giant organized a conference around artificial intelligence on February 8, in Paris. It was a very dense week on AI, because in addition to the formalization of Bard and the event in the capital, Microsoft made announcements with ChatGPT and OpenAI on February 7.
Bard’s existence had been aired by the press a few days earlier. Indeed, the media had been able, from the end of January, to observe the operation of the chatbot. Its final name was already practically found – it was referred to as “Apprentice Bard” – and differences with ChatGPT had been observed, in particular on the freshness of the information.
This week, however, turned into a nightmare for Google. On February 6, a demonstration by Bard showed a factual error, when it could have been checked beforehand (a discovery in astronomy was wrongly attributed to the James Webb telescope). And on February 8, the conference said very little about Bard, with Google focusing on other things.
The event left a taste of unfinished business for many observers, because the Mountain View firm seemed to lack enthusiasm for Bard. By contrast, Microsoft’s announcements with ChatGPT were otherwise spectacular. These three factors (Bard’s mistake, weak conference, dynamic rival) caused Google’s share price to plummet, wiping out $100 billion.
How does Google Bard work?
Bard relies on the LaMDA language model, an acronym for Language Model for Dialogue Applications. Designed by Google, it was presented to the public for the very first time in 2021 – since then a new, more complete version has been developed. LaMDA mobilizes an artificial neural network, Transformer, developed in 2017, and on AI research dating back to 2020.
Bard, whom Google describes as a ” experimental conversational AI service », is not yet accessible to the general public — its existence nevertheless gives meaning to the work on the LaMDA language model and the Transformer artificial neural network, by proposing a concrete application. Bard is also connected to the Internet, to have recent data.
Google clarified that its current version of Bard runs a lightweight version of LaMDA. ” This model, which is smaller and requires less computing power, allows us to deploy to a greater number of users and process a larger volume of feedback. “Explained the American company, when presenting its conversational agent.
Why did Google reveal Bard?
Ever since introducing ChatGPT in November 2022, Google has come across as being on the defensive and lagging behind on content-generating artificial intelligences. This sentiment is exacerbated by the immense public and media interest in OpenAI’s chatbot, since accentuated by Microsoft’s massive investment in OpenAI and all past announcements.
The feeling that Bard was announced hastily, in reaction to ChatGPT, is reinforced by the fact that Google has said little about LaMDA (the language model on which its chatbot is based) and its AI and chatbot projects. Admittedly, LaMDA was unveiled in May 2021, and refreshed the following year. But the web giant does not seem in a hurry.
Added to this are the echoes and analysis of the media.
We know that Microsoft sees in ChatGPT an extraordinary lever to “kill” Google on the search market (or, at least, to weaken it). THE New York Times mentioned a sort of “red alert” at Google since the release of ChatGPT. It has even been reported that the two founders, Larry Page and Sergey Brin, were called back urgently. In short, there is concern in the air.
The next annual Google I/O conference, expected this spring, could be an opportunity for Google to pull out the heavy artillery, in order to regain some of the initiative – and the media spotlight. Corridor rumors suggest that the net giant would have on the rails about twenty products and services embedding artificial intelligence.
Why Google fears ChatGPT, and doesn’t want to go too far with Bard?
Microsoft, today, has almost no position in online search — its in-house engine, Bing, occupies only a few percent in the world. For the software giant, this activity is not strategic. In any case, not as much as for Google. However, chatbots could revolutionize the way Internet users access information.
To put it quickly, there is a possibility that in the more or less distant future, search engines will become obsolete — we would just ask a chatbot to find us the information. But today, Google exercises its dominance on the web, because it has become the central gateway for the majority of Internet users. And it is with her that he earns money, through advertising.
Microsoft has nothing to lose. This is a unique opportunity, both to disrupt the web with ChatGPT and to gain the upper hand on Google. It would be a first in twenty years. We therefore understand why Google is trying to prevent the revolution that is ChatGPT. It is its dominance over research that is in jeopardy and perhaps even its long-term existence.
In short, the war of artificial intelligence — in any case chatbots, Bard vs ChatGPT and Microsoft vs Google — has barely begun. The hostilities are just starting and we don’t know when and how this conflict will end, if there will be other belligerents (like Amazon or Apple?), nor what the victims will be. And on the web side, there may be collateral damage.
Subscribe to Numerama on Google News so you don’t miss any info!
