‘Bard,’ Google’s AI chatbot, was released three months after the tech giant issued a “code red” in response to ChatGPT’s rising popularity.
Google’s Bard is the company’s response to Microsoft-backed OpenAI’s ChatGPT, an AI chatbot that gained notoriety late last year for its ability to quickly produce conversational writing that sounds human.
Bard will continually use data from the Internet to generate replies, unlike its rival ChatGPT, which only uses information from the sizable corpus of text it was trained on.
Google CEO Sundar Pichai argues that Bard can deliver up-to-date responses by doing so.
Contrarily, because ChatGPT’s training set concluded with data from 2021, its responses need to give specifics regarding recent news and research. As a result, it needs to provide current information, such as statistics or analyses based on President Joe Biden’s State of the Union address.
Bard leverages Linguistic Model for Dialogue Applications (LaMDA), a language system that Google created two years before the AI chatbot craze began, in addition to having a whole separate data source.
Surprisingly, the Transformer neural architecture is the foundation for several AI chatbots, including ChatGPT’s third generation of generative pre-trained Transformers. LaMDA is constructed utilizing this neural architecture (GPT-3).
LaMDA, in contrast to GPT-3, is trained with 1.56T words and 137B model parameters.
Additionally, it adheres to Google’s standards for quality, safety, and groundedness to ensure the generated responses are high-quality and safe.
In the interim, Google will scale the AI chatbot to more people using a lighter, more compact version of LaMDA that requires fewer processing resources. Google will also gather comments.
As of this writing, Bard is only accessible to a set of beta testers. Google promises it will be accessible to the general public in the upcoming weeks, despite having already missed its first projected launch date of late February.