It’s early days for Google’s AI tool Bard. Developed to combat the popularity of ChatGPT, the new AI wonder kid, Bard is still in its nascent stage. The Google execs are perfectly aware that the tool is far from infallible when it comes to answering queries.

According to a CNBC report, the management has now entrusted the staff with the duty of fixing the wrong answers. A leaked email from Prabhakar Raghavan, Google’s vice president for search, has revealed an ardent request to the workers to improve the tool, along with an extensive list of do’s and don’ts. Raghavan wrote, “This is exciting technology but still in its early days. We feel a great responsibility to get it right, and your participation in the dogfood will help accelerate the model’s training and test its load capacity (not to mention, trying out Bard is actually quite fun!).”

‘Dogfood’ or ‘dogfooding’ is a corporate buzzword, referring to the practice of testing your own products before making them available to the public.

The staff has been instructed to write answers on topics that fall within their expertise. “Bard learns best by example, so taking the time to rewrite a response thoughtfully will go a long way in helping us to improve the mode,” the document says. In order to make the tool the number-one source of accurate answers, the Google bosses have rolled out a list of do’s and dont’s.

The ‘Do’s:

Write What You Know
Employees have been instructed to stick to what they know best when writing answers.

Remain Polite
The staff has been asked to mind their tone while answering. “Keep responses polite, casual, and approachable,” has been the brief.

Stick To I, Me, Us
Answers have to be in the first person so that users feel they are interacting with a ‘real’ person.

Be Neutral
The answers have to be objective and impersonal, with no hint of personal bias or partiality. The employees have also been told to downvote answers that offer legal, financial, or medical advice or come across as “hateful or abusive.”

The Don’ts

Never Assume
The search engine behemoth has asked its employees to steer clear of stereotyping based on race, religion, gender, etc. in their answers.

Don’t Personify Bard
The employees have to be careful not to attribute human-like characteristics like emotion to the Bard.

Don’t Rewrite
“Don’t try to rewrite it; our team will take it from there,” the document states.

The management has come up with a recognition policy to incentivise employees. Workers who agree to help out with improving Bard and provide feedback will be rewarded with a ‘Moms Badge’, which will be displayed on their profiles.