What you need to know
- Google launched Bard shortly after Microsoft integrated ChatGPT into Bing and Edge.
- A former Google engineer claimed that Bard was trained using data from ChatGPT.
- Google denies using ChatGPT to train chatbots.
As if Google and Microsoft had no room to fight, the battle between AI chatbots is only heating up. Following ChatGPT’s popularity and Microsoft’s integration into his Bing, the former employee says Google is using data from his OpenAI chatbot to train its own competitor, Bard. claimed.
according to the information (opens in new tab), former AI researcher Jacob Devlin allegedly resigned after expressing concerns to Alphabet CEO Sundar Pichai and other executives about how Bird was trained. He apparently believed the team behind Bard was using his ShareGPT, a platform where users post their interactions with his OpenAI chatbot.
Devlin was concerned that not only would using the data violate OpenAI’s terms of service, but Bard’s response would look too much like ChatGPT. A source told The Information that Google stopped using ChatGPT for training after his Bard raised concerns.
In a statement to The Verge, a Google spokesperson said, “Bard is not trained on data from ShareGPT or ChatGPT.”
Google has claimed that Bard is built using LaMDA. LaMDA is the language model the company has been working on to enable more conversational AI (at one point Google engineers thought it was sentient). Still, Bard’s announcement and eventual public testing seemed like a pretty reactive move from Google. That would make Devlin’s claim even more likely.
As for Devlin, The Information indicates that he will leave Google to work with many other AI researchers at OpenAI.
Meanwhile, Google appears to be ramping up its efforts to make Bard compete with ChatGPT. The effort is reportedly called “Gemini.” Of course, with ChatGPT now being integrated into Bing and the resulting increased use of the Microsoft search engine, it makes sense that Google would use all the tools at its disposal to sniff out Bard.
