What is the issue
According to The Information, OpenAI’s success “forced” Google parent Alphabet’s two AI research teams – Google Brain AI and DeepMind AI lab – “to overcome years of intense rivalry to work together.”
Citing two people with knowledge of the project, the report said that the two teams are working together to develop software, which is internally called Gemini, to compete with OpenAI.
Moreover, the report also claimed that Google used data from OpenAI’s ChatGPT, scraped from a website called ShareGPT, to train Bard. ShareGPT is a website where people publish conversations they have had with ChatGPT.
A former Google AI researcher, who had raised red flags about its development to company CEO Sundar Pichai and other executives, reportedly spoke against using that data.
Here’s what Google has to say
Google has denied that ChatGPT data was used to train Bard. “Bard is not trained on any data from ShareGPT or ChatGPT,” The Verge quoted Google spokesperson Chris Pappas as saying.
Google Bard has guardrails
Recently, The Wall Street Journal reported that “Google made Bard boring on purpose.” When asked about Google’s take on this, Sissie Hsiao, a Google vice president who oversees Bard, reportedly said that chatbot is an “early experiment” and the company wants Bard to “output things that are aligned to human values.”
“We feel really good that Bard is being safe and actually people are finding those guardrails,” Hsiao was quoted as saying. She also said Bard is trained on publicly available information. It is to be noted that ShareGPT’s data is also available for people to read.
