“Do not include information that identifies others or yourself in your conversations with Bard.” The guide is from the Bard Data Privacy Help Center. Accessible in France since Thursday, Google’s artificial intelligence solution, intended as a competitor to ChatGPT, also warns people directly before chatting with the AI model.
So, to better train its language model using artificial intelligence, Google turns to humans. Their role is to review and note requests made to Bard.
For up to 72 hours
Google ensures that these conversations are anonymized before being transmitted to human teams, in particular, by unlinking the Google account associated with the responses. But what is striking is that it is not possible to withdraw from the process.
Basically, the history of exchanges with the AI model is kept for 18 months, indicates the conditions of use of Bard. This duration can be modified with the possibility of opting for an automatic withdrawal of 3 or 36 months.
Also, it is possible not to record your trading history with the “Bard Activity” option. By disabling this option, AI conversations will not be saved to your Google account. But they will still be accessible to human reviewers.
“Even when Bard activity is disabled, your conversations are saved in your account for a maximum of 72 hours,” describes the company, which advances a need for context in exchanges with its AI as well as Bard’s security and protection measures. .
Google claims that the use of staff improves its technology “to make Bard safer and more useful.” Human checks are made from a random sampling of conversations with Bard.
Source: BFM TV
