Google launched its conversation technology Bard, a rival to Microsoft-backed ChatGPT, last week, and it didn’t go as smoothly as the tech giant would’ve hoped.

Google employees called the unveiling “rushed” and “botched,” and the Alphabet stock price fell nearly 9%. Now, the company is asking its staff to address its artificial intelligence search tool’s inaccuracy — by rewriting Bard’s answers themselves, CNBC reported.

Related: Google Gears Up to Compete With Microsoft-Backed ChatGPT

In a blog post introducing Bard, Google CEO Sundar Pichai said the company’s chatbot “draws on information from the web to provide fresh, high-quality responses,” suggesting that it can provide information about recent events (ChatGPT struggles with that, according to The Verge).

Google’s vice president for search Prabhakar Raghavan asked staffers on Wednesday to help the company ensure Bard provides accurate responses, per an email viewed by CNBC. The message included a dos and don’ts page with instructions on how to fix the ChatGPT competitor’s mistakes.

“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.

Related: ChatGPT: What Is It and How Does It Work?

Google’s dos include keeping responses “polite, casual and approachable” and “in first person” while maintaining an “unopinionated, neutral tone.”

The don’ts? “Avoid making presumptions based on race, nationality, gender, age, religion, sexual orientation, political ideology, location, or similar categories” and refrain from referring to Bard as a person or implying emotion.