Google Bard: the alternative to Chat GPT

It seems that some alarm has gone off in Mountain View offices. The popularity that Open AI is achieving with its conversational chat operated by Artificial Intelligence, as well as the uses that are beginning to be announced, have meant that Google has had to accelerate its plans at the risk of falling behind in its strongest point: innovation.

 

There is still a lot of uncertainty, but we are going to tell you everything that is known about Google Bard gate below.

Being somewhat reductionist, we could summarize it as the competition of Chat GPT by the dominant search engine of the market, Google: a full-fledged Google killer.

 

But we do not  like to stay in the big gimmicky phrases, so let's go a little more to the technical side.

 

We will say that it is an automated conversational chat, with an interface based on natural language models, which allows the user to interact with the software by asking complex questions and getting specific answers.

Yes, it is, in essence, a chatbot. What happens is that it has been trained based on machine learning and has been given - this is important - access to the Internet to generate its answers with an updated and almost unlimited source of knowledge.

 

For Google, as of today at least, it is going to be a new search feature.

One of the things we are more surprised about this announcement - apart from the suddenness - is that a new naming has been used, when it has been known for years that there is a previous program under development.

 

The name of this previous tool is LaMDA, which stands for Language Model for Dialogue Applications.

 We understand that Google's alternative to Chat GPT has been Bard because they do not fully trust the maturity of LaMDA, and because, in a way, they do not want to tarnish the perception of the latter language model, in which they have invested so many years and resources.

 As this epigraph titles, LaMDA is the basis of Bard, but also a source of controversy within Google, which came to public attention quite some time ago.

 

The reason was that one of the engineers in charge of its development and training had such a deep and philosophical conversation that he came to the conclusion that this AI could have developed a certain self-awareness.

 

Blake Lemoine, that was the name of the Google worker, was not only disavowed: he was also suspended without pay.

 

This is not the purpose of the article, so we will not go into the ethical aspect any further, but if you want to know more, you can read the transcript of this conversation here.

But let's go back to the present and see what Google has announced with Bard and how it differs from Chat GPT, and also what its real competitive differential advantage is.

 

The most important thing is that, unlike its rival, Google promises to integrate this chat with the search engine without having an external interface and, what is more interesting, to provide it with an Internet connection, so that its responses will gain in immediacy and timeliness, as we pointed out before.

 

Evidently this is a defensive move, since, with a multi-million dollar investment, one of Open AI's partners (the creators of Chat GPT) has become Microsoft, and there is already an announcement that the company is preparing the integration of Chat GPT 4 with Bing.

 

This could represent a huge blow to Google and cause it to lose its dominant position, jeopardizing one of the company's most important sources of revenue: Google Ads and its ads.

Nor have they been very precise with the dates, the statement only speaks of "in the coming weeks".

 

What they have made clear is that it will not have its own domain or interface: it will simply be added to that of the search engine, thus concentrating the experience (and traffic).

We are facing a new twist, and it seems that it will be truly transformative.

 

Although we have been seeing search engine algorithms use natural language for years, the use of conversational chats is going to "kill" SERP and, in a way, keywords as well.

 

This is going to be especially hard for informational searches, since, with this model, there is only one answer per query. There is no point in thinking about rankings and positions anymore.

 

The first examples they have shown seem promising, but some features have also been exposed that have not been entirely to their liking. It is especially criticized that the answers do not cite the sources, with which, basically, it would be feeding from billions of web pages and contents, without providing credits.

Images | Front page generated with MidJourney, Unsplash and linked article from Google.