Bard, Google's new intelligent chat

Contenido:

Google wants to remain the Internet search leader, and has been developing an AI for years. Currently it is about to see the light of day to compete directly with Chat GPT. In today’s post we tell you more about this new artificial intelligence language model.

What is Google Brad?

Bard is the conversational AI that Google has developed based on its own language model known as LaMDA, Language Model for Dialogue Applications.

Google and Alphabet CEO Sundar Pichai has written an article on the progress of their AI and in it he highlights that their research project called Transformer and the paper published in 2017, is the basis for many of the generative AI applications we are starting to see today.

How is Bard going to work?

The first version to be launched will be lighter, in terms of the computational power needed to scale it to more users and receive more feedback.

Google has already used AIs to improve its search engine searches. Initially they used BERT, Bidirectional Encoder Representations from Transformers, whose main novelty was to be able to understand the meaning of previous and subsequent words. Thus, with this system, it was possible to perform text classification tasks, answer questions and extract information.

However, two years ago this system was replaced by MUM, Multilingual Unified Model, a model 1000 times more powerful than its predecessor. This pre-trained multilingual model can be tuned for a wide variety of natural language tasks in several languages.

How is Bard different from Chat GPT?

One of the main differences between these two AI models is that Bard will be directly connected to the web. This will allow it to give us information or summaries on current topics, something that Chat GPT cannot do at the moment. Chat GPT’s database is limited to September 2021.

When will it be available?

But when will the new Brad model be available? According to Google’s CEO, on February 6, they opened the system to their trusted testers in order to make it available to the general public ‘in the coming weeks’.

It is also known that during the month of February they are going to start testing with their API, with developers, creators and companies. Its main objective is to have a suite of tools and APIs to make it easy to implement this AI in new applications.

In addition, Google wants to make sure that the release of its AI is safe and does not damage its reputation with potential bugs. Pichai and his colleague Jeff Dean, head of artificial intelligence at Google, have stated that Google must act more conservatively than a startup because of the reputational risks associated with the current state of AI language models.