Sundar Pichai Will Respect Google Employees If They Trial Bard Two To Four Hours A DAY

Google asks all its employees to spend two to four hours a day testing Bard. (photo: dock. google)

JAKARTA - In order to ensure that Bard's new Artificial Intelligence (AI), Google asks all its employees to spend two to four hours a day testing the ChatGPT-like chatbot.

In a leaked company email, Google CEO Sundar Pichai wants his employees to make sure the upcoming Bard is functioning as best they can.

Bard last week was launched in a bid to beat Microsoft, which introduced its own AI chatbot into the Bing search engine.

However, what came as a surprise Bard featured a false statement about the James Webb Space Telescope, where the chatbot offered three points in the affirmative.

That includes stating it was the Webb Telescope that managed to take the first image of a planet outside the Solar System. Whereas the first image of the exoplanet was actually taken in 2004 as stated by NASA. With this error, it caused the company to lose 100 million US dollars in stock prices.

Furthermore, Pichai also said, although there have been thousands of Google workers testing Bard over the past few days, he said he would appreciate it if all of its employees contributed more deeply to testing the chatbot.

"(Companies) often seek input from Googlers to help make our products better," a Google spokesperson said in response to the leaked email to Gizmodo, quoted Thursday, February 16.

Pichai did not explain whether the two to four hours would be valid every day or divided over a longer period of time.

It is not known for sure whether all Googlers (as Google employees) around the world have received the same request. This is because the company has fired 12,000 of its employees globally.

Even so, the parent company Alphabet, still employs more than 170,000 people worldwide. In the email, it was also revealed that Google had started internal testing, called dogfooding.

The testers are reportedly investigating quality and safety issues with AI searches, along with groundness, which may be linked to whether AI-generated text answers are read human.