“Bard was not trained on any data from ShareGPT or ChatGPT,” a Google spokesperson said.
Updated – Thu, 3/30/23 at 10:33am
New Delhi: Google has denied reports that it is copying Microsoft-owned OpenAI’s ChatGPT to train its AI chatbot called Bard.
A report in The Information said that OpenAI’s success “forced two AI research groups within Google’s parent company Alphabet to overcome years of fierce competition to join forces”.
The report, citing sources, said software engineers from Google’s Brain AI group were working with employees from DeepMind, Alphabet’s sister company, to develop software to rival OpenAI.
“The joint effort, known internally as Gemini, began in recent weeks as Google’s first attempt to compete with OpenAI’s chatbot following setbacks with Bard,” the report said.
However, a Google spokesperson told The Verge that “Bard was not trained on any data from ShareGPT or ChatGPT”.
At the same time, Google announced that it would open up access to its ChatGPT competitor “Bard” as an early experiment in user-generated AI collaboration.
Bard is already available in early access in the US and UK, and the company says it will expand access to more countries and languages over time.
Bard, like OpenAI’s ChatGPT and Microsoft’s Bing chatbot, is based on large language models (LLMs), specifically a light-weight optimized version of LaMDA, which the tech giant says will be updated with newer, more powerful models in the future.
Users can interact with the Bard by asking questions and refining their answers with follow-up questions.
