With Google Bard failing miserably on its launch, the company is now asking its employees to train their AI chatbot as much as possible.
In an email sent to the employees, Google’s VP of Search Prabhakar Raghavan asked people to rewrite Bard’s responses on well-known topics. Hoping that training it with factual answers makes it better, he also mentioned a bunch of Dos and Don’ts for employees training the Bard.
Bettering Google Bard With Personal Training
As you know, last week’s launch of Google Bard was a great mess – denting the company’s image and wiping out around $100 billion from its valuation! All because the AI chatbot gave false information when asked a question regarding the James Webb Space Telescope.
As it stirred many in the community, including some of Google’s own staff, the company is now asking its employees to better it by training personally! In an email sent to all the employees this week, Google’s VP of Search Prabhakar Raghavan asked employees to rewrite Bard’s responses on topics they know well.
Saying that the chatbot “learns best by example,” Raghavan said training Bard with factual answers would help in improving its accuracy. And to do that in an organizational style, he listed a bunch of “dos” and “don’t” for the training process.
Some of these include curating the responses to be in first person POV, being unopinionated and neutral, polite, casual, and in an approachable tone. Employees are also told to “avoid making presumptions based on race, nationality, gender, age, religion, sexual orientation, political ideology, location, or similar categories.”
Aside from this, employees are instructed to downvote any answers put up by Bard containing “legal, medical, financial advice” or are hateful and abusive. This comes days after Google CEO Sundar Pichai asked the employees to do the same – wanting them to spend a few hours each week testing the AI chatbot.