Spend four hours with AI chatbot Bard, rewrite bad responses: Sundar Pichai tells Google employees

In an email to employees, he added the company will send detailed instructions next week, according to a CNBC report.

Meanwhile, Prabhakar Raghavan, Google’s vice president for search, asked staffers in an email on Wednesday to help the company make sure Bard gets answers right, according to the report.

The email included a link to the do’s and don’ts page with instructions on how employees should fix responses as they test Bard internally.

Also read: ChatGPT, Bard & Ernie: The three musketeers of AI

Staffers are encouraged to rewrite answers on topics they understand well.

Discover the stories of your interest

“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 said.

Google also instructed employees to give a “thumbs down” to answers that offer “legal, medical, financial advice” or are hateful and abusive.

Raghavan said Google will invite the top 10 rewrite contributors from the Knowledge and Information organization, which he oversees, to a listening session.

Also read: ETtech Explainer: Big Tech battle it out for AI control

There they can “share their feedback live” to him and people working on Bard.

This comes after the tech giant unveiled Bard earlier this month. However the generative AI chatbot made a factual error on a query during a promotional video after which shares in Google’s parent company Alphabet fell 7.7%, wiping $100 billion off its market value.

Also read: How one mistake by Google’s AI chatbot Bard wiped off $100 billion in market cap for Alphabet

Following this, even Google employees criticised Pichai, describing the rollout as “rushed,” and “botched”.

Bard’s inaccuracy highlights the challenge for Google as it looks to take on OpenAI’s popular AI chatbot ChatGPT which Microsoft is integrating into its search engine Bing.

Microsoft on Wednesday said more than 1 million people signed up on the waitlist to try out the new Bing Search with an integrated ChatGPT functionality in just 48 hours.

Chatbots like ChatGPT and Bard are based on the concept of a generative AI. Which means that the platform is trained on a massive amount of available information on any given topic online to generate responses for a query.

Also read: ETtech Explainer: warnings aplenty against AI chatbot biases, ignore them at your own peril

Meanwhile, Pichai, announced in a blog post that the company was opening Bard to a group of trusted testers, ahead of making it more widely available to the public in coming weeks. In his letter, Pichai told employees that Google has not always been the first to release a product, but that hasn’t hampered its ability to win.

“Some of our most successful products were not first to market. They gained momentum because they solved important user needs and were built on deep technical insights,” CNBC quoted Pichai as saying in an email.

He added that the most important thing the company can do right now is “to focus on building a great product and developing it responsibly,”

He also added that the company has thousands of external and internal people testing Bard’s responses “for quality, safety, and groundedness in real-world information.”

“Channel the energy and excitement of the moment into our products,” Pichai wrote. “Pressure test Bard and make the product better.”