.
War has been declared!
No, not the ones in the history books, but the war between the Big Techs, more specifically those involved in the first and second places of search engines, or if you prefer: Google and Bing, respectively!
Last Monday (06/02), Sundar Pichai, CEO of Alphabet’s Google, announced at the company blogits new product in the field of Artificial Intelligence and which promises to challenge the one that until then had the spotlight on it – ChatGPT.
Your opponent’s name is Bard!
What is Bard?
Bard is what Google has been calling the experimental AI-based conversational service, which was announced on the company’s blog and will initially be accessible only to “trusted testers” before making it available to the general public.
For those who have already heard or even become familiar with ChatGPT, it promises to be your direct “competitor”.
For those who don’t, it’s a system based on technologies and studies, involving deep learning, LaMDA (Language Model for Dialogue Applications) and which means Language Model for Dialogue Applications, developed at Google and others that are internally called PaLM, Imagen and MusicLM.
Despite this bunch of acronyms and in concrete terms, Bard can appear under an application that, subjected to more complex and elaborate questions in the same language as we use in everyday life, is able to understand it and give a practical, useful and comprehensible answer. , also in natural language.
There is still no precise and detailed information on its operation, but it is imagined that, like the tool with which it intends to rival, it is a Generative Artificial Intelligence.
What is generative artificial intelligence?
Like human intelligence, it is the intelligence that is creative, that is, that generates authentic results, not just copying or transcribing snippets of information and data collected from the training to which it was submitted, but responses entirely produced by it. If you prefer, original content.
The cited researches that are included in the project (PaLM, Imagen and MusicLM), allow this AI to involve more than text or words, but also images, videos and audio.
Therefore, even if we don’t have details yet, it is assumed that Bard is more “talented” than ChatGPT, which so far only produces content in the form of text.
Why Bard?
The translation of Bard into Portuguese means bard, which was the name given to the teller of stories, fables, myths, legends, sometimes sung and sometimes also had the ability to compose verses and recite poems, or even knew literature and had the gift of reading, this in ancient Europe (15th century) and, therefore, at a time when reading was something reserved for very few.
William Shakespeare, who was a playwright, poet and actor, was considered a bard.
That is, the bard was someone who had the function of entertaining the nobility, but also had an educational role and of spreading knowledge and providing culture.
ChatGPT vs Bard
It is not official, however, the repercussion and demonstrated potential of ChatGPT and the many associations, especially with Microsoft that has already invested heavily and promises to invest another 10 billion dollars and should include it in Microsoft Bing, made Google anticipate the announcement.
It is already known that the Mountain View company had been dedicating itself since 2018 to different studies focusing on AI, as well as investments in companies in the sector, such as the startup specialized in AI created by some former leaders of OpenAI (creator of the ChatGPT), the anthropogenic.
However, despite the great dedication that had been given to the subject and gradually having incorporated some of the results of this work, as in the case of the Bert update and later, the MUMas Pichai wrote, the idea was for the service to deliver “high level of quality, security and grounding in real-world information”.
This is reinforced by saying that the improvement of the service will be the result of combining external feedback – the so-called reliable testers – with the company’s internal tests.
In short, unlike other products, which are traditionally released only when high levels of reliability and maturity are reached, it seems that Bard will still need some adjustments until it reaches the desired and usual level.
Another point that shows the clash is the information that initially the model that will be made available is much smaller, requiring little computational demand and thus allowing access by more users and producing more feedback.
It even seems like a poke at the rival, since there have been constant reports that ChatGPT is completely inaccessible at various times of the day, due to the high demand of users, but with MS behind, who knows, who knows soon it will run in the cloud from the company of Bill Gates.
The same happens on Google’s side, which has gigantic Cloud Computing infrastructure, which once used, can make Bard a troublesome competitor.
What changes in searches with Bard?
An announcement with more details is scheduled for Wednesday – the same day this post was published – and as such, we still don’t have much to know about it.
However, it is certain that at some point Bard will be integrated into searches.
From what is possible to conclude and from what already exists, for some research, both the way of doing it and the result change a lot and may indicate a paradigm shift for users and content producers for the web, a subject that we even address in the article “What is the future of Artificial Intelligence?”.
In practical terms, imagine that you intend to make the road to Santiago de Compostela by bicycle in August 2023 and leave São Paulo, Brazil.
Finding out what needs to be planned in terms of logistics, equipment, transfers, accommodation, in short everything that involves a trip of this type, would require possibly more than a hundred different searches in “traditional Google”, as well as access to several of the websites that appear on the SERPs.
With a tool like Bard, or Microsoft Bing with GPT-4, a possible search would be:
“Tell me everything I need to plan a trip leaving São Paulo, Brazil, going by plane, in August 2023, to make the road to Santiago de Compostela by bicycle (cyclotourism), including accommodation, maintenance infrastructure for the bike , food points, maps, items I need to bring, costs and visas”.
Theoretically, both one and the other will be able to answer questions with this level of detail, which in the current model, as we said, is only possible by carrying out numerous individual surveys.
Even a hypothetical search like the one suggested can be further refined and the AI also redoes the answer until it is considered satisfactory.
The numerous tests that are possible to find with situations of this kind using ChatGPT, show that there are still several aspects to be improved, but many of the results are impressive.
It is known that the Google service was based on a much higher number of parameters for training and is more up to date, since the “competitor” was trained with data “only” until 2021, so that nothing that happened in 2022, he is able to answer.,
Whether in one case or the other, what we’ll see shortly will be just the beginning, but it should be very promising, above all, giving users broad, practical and quick answers.
The company’s main and oldest tool, represented by the search engine, was gradually changing in this direction and for many searches, instead of just listing a bunch of links, it displays direct answers to certain questions, such as currency quotation, conversion of metric units, in the translator, or even in featured snippets and rich snippets.
With Bard, this type of delivery of results is expected to be even more highlighted.
The Bard for Developers
Google’s own CEO has already given information that other AI-based services should emerge.
In the same release, Sundar Pichai mentioned that they intend to release a set of tools and APIs to the industry, for creating tools based on artificial intelligence.
It means that the company is keeping an eye on developments in technology, so that companies have different answers and solutions, based on data that they themselves produce, but based on the AI models of the search giant.
In this way, a new era of data-driven applications will produce much more strategic and predictive assessments. The level of automation of processes and procedures, as well as continuous improvements, must also be significantly boosted.
Intelligent virtual assistants, which are widely used today, could evolve to a much higher level. Siri, Cortana, and Alexa can and should get a lot more capable.
Envisioning all the areas and benefits of this new model is even difficult given the range of possibilities.
However, as we have already discussed, it is necessary to look at the other end and establish ethical limits of artificial intelligence, as well as “not so noble” uses or with conflicting interests.
Conclusion
Through Bard, Google is once again using artificial intelligence to provide information and transform it into useful and efficient knowledge.
.
