ChatGPT Rival Bard AI Accused Of Using Gmail Datasets; Google Responds

Google AI bot stirs controversy when it says it is using Gmail dataset for training purpose.

New Delhi: Google seems in hurry to enter into AI race when its competitors ChatGPT and Bing are already cementing their roots in the race of domination. On that backdrop, it unveiled its AI bot a while back and now announced to expand it for public use in UK and US. Google faced dire backlash from netizens at the unveiling of AI bot Bard when it answered wrongly to a question. Now, as Google has announced to open its bot Bard for US and UK users in a limited capacity, it has again fallen into a controversy with allegations of using Gmail dataset for training purpose.

A screengarb appeared to show, which has been shared on Twitter, that when asked Bard about its dataset origin, it replied several sources including Google’s internal data from Google search, Gmail, and other products. The other sources, it mentioned, are publicly available datasets include datasets of text and code from the web such as Wikipedia, GitHub, and Stack Overflow and Data from third-party companies including those which have partnered with Google to provide data for Bard’s training.

Umm, anyone a little concerned that Bard is saying its training dataset includes... Gmail?

I'm assuming that's flat out wrong, otherwise Google is crossing some serious legal boundaries. pic.twitter.com/0muhrFeZEA — Kate Crawford (@katecrawford) March 21, 2023

However, the company responded on the allegations and said this is an error and Bard’s dataset isn’t using GMAIL data. Google workspace tweeted, “Bard is an early experiment based on Large Language Models and will make mistakes. It is not trained on Gmail data.”

Twitter user Kate Crawford, an AI researcher, who put the screengrab, commented further asked if she could confirm that was a Bard hallucination and that there’s no Gmail data included whatsoever in the training process.

Another user named Simon Bisson shared a screenshot in which she asked the same question. Bard told her that it had been trained on a massive dataset of emails from real people including friends, family, colleagues and business associates…the source of the emails it had been trained on is confidential.

This is what it told me: pic.twitter.com/ugJBG694YZ — Simon Bisson (@sbisson) March 21, 2023

Google seems in hurry after its competitors Microsoft and OpenAI have already launched their AI bot Bing and ChatGPT respectively in the market for public use.