Google’s advanced AI chatbot Bard AI has been accused of using ChatGPT data to train its AI model. The allegations came to light when the former Google AI researcher shared his concerns regarding the usage of rival ChatGPT data and the possibility of chatbots generating responses that might turn out similar.
Although Google has denied all the accusations claiming the usage of ChatGPT data to train Google AI chatbot Bard stating, “Bard is not trained using any data from ChatGPT or ShareGPT”.
Key points:
- Jacob Devlin, a former Google AI engineer left Google and joined its rival OpenAI immediately after warning Google to not use OpenAI Chatbot ChatGPTs data, which might lead to a violation of OpenAI’s terms of service.
- Google denies all the accusations claiming Bard has been trained on ChatGPT data.
- Some rumors claim Google has ordered the DeepMind division to generate a new collaboration with the brain team.
Is Bard trained on ChatGPT data?
Google sent a statement to WinFuture saying that Bard’s training did not rely on data from ChatGPT. “Bard is not trained on data from ShareGPT or ChatGPT,” the company told the publication.
Google Bard is an AI chatbot, and like any other generative AI model, it is produced using a combination of machine learning models and training data. Google Bard is trained using LaMDA (Language Model of Dialogue Application) as its machine language model. Although it’s for the syntax of the chatbot. When it comes to training data, a variety of open-source libraries are free for AI researchers.
According to reports, there is a high possibility Google might not have scoured these data through them, as the public domain data might not be enough to come up with a generative model like Bard. Instead, there is a chance Google might have trained on ChatGPT data shared by OpenAI publicly through ShareGPT.
The data includes outputs of ChatGPT at the core, along with the accessibility of sharing the responses on the website from the chatbot of OpenAI. These allegations haven’t just arisen out of nowhere, Jason Devlin, a former Google AI researcher, apparently left Google after raising a warning regarding the top excess which utilizes the data from OpenAI’s ChatGPT, which can result in violation of OpenAI’s terms of service.
NEW: Prominent Google AI researcher resigned after warning Alphabet CEO Sundar Pichai and other senior execs that Bard—Google’s rival to ChatGPT—was *using data from ChatGPT*.
— Amir Efrati (@amir) March 29, 2023
Big no-no in that world. https://t.co/a5NeclJPK5 w/ @jon_victor_ pic.twitter.com/YEZqEqpzPS
The former Google AI researcher is currently working at OpenAI. According to reports, Devlin was also concerned that if google tapped into the data from OpenAI’s chatbot, then bard would provide simple outputs like ChatGPT. In addition, there are further claims that state Google ordered the DeepMind division to create a new collaboration with the brain team. The primary motive for creating this collaboration was to start a new initiative known as “Gemini” as an internal name.
Google’s response to the accusations
Google has denied all the accusations claiming the AI chatbot developed by Google has been trained on ChatGPT Data. To be precise, Google’s spokesperson, Chris Pappas stated, “Bard is not trained using any data from ChatGPT or ShareGPT”.
Previously, some reports claimed Google’s CEO Sundar Pichai had asked employees to take extra hours to iron out a few kinks in the ChatGPT competitor.
For the moment, it’s unclear to state whether Google really used data from ChatGPT to train Bard or whether it’s just rumor. Although there is a chance google might have relied on ChatGPT data before Bard’s launch. However, whether Bard is still operating on any of OpenAI’s training data is unclear. As Google has also denied all the accusations that came their way.
Google Bard has been battling ethical concerns and a lack of progress. Even Google’s employee feels Bard was rushed by advancing its launch timeline so it can compete with ChatGPT which turned out to seem botched.