Google claims that it didn’t train Bard on your private emails

In a recent statement, the tech company Google confirmed that its chatbot is not trained on text from private Gmail accounts. In response to ChatGPT, Google has introduced its own internet web search chatbot i.e., Bard.

The matter came to light when an AI researcher questioned Bard regarding its training. Surprisingly Bard responded to the question with an answer that mentioned the internal data from Gmail. Previously, we have come across the news that Google fired one of its employees (Blake Lamoine) for sharing the company’s information. Blake claimed that it is trained on text from Gmail.

So, what should be done in such a confusing situation? Well, as per the information (The Register) questioned Google about this. In response, a representative said, similar to all LLMs, Bard has the potential to generate accurate responses or responses that present misleading information. Such a response regarding the training of Bard from Gmail is an example of that. The company doesn’t make use of the personal data of users from Gmail or any other personal app or service to update Bard.

Do note that Google has recently introduced Bard. The company has invited users from the UK and the US to join the waitlist for using Bard. In contrast to earlier tests of Microsoft’s Bing, Bard doesn’t seem to produce random and erratic text. It is still prone to making up misleading information and can still be persuaded to respond to inappropriate demands.