Our AI writing assistant, WriteUp, can assist you in easily writing any text. Click here to experience its capabilities.

Google asks employees to rewrite Bard's bad responses, says the A.I. 'learns best by example'

View Original View Raw

Summary

Google has asked its employees to help fix bad responses from the company's AI search tool, Bard. Employees are instructed to rewrite responses thoughtfully, in a polite and neutral tone, avoiding stereotypes and offering no legal, medical, or financial advice. As an incentive, contributors will earn an internal badge and a chance to meet with the Bard team. Google leaders are hoping that by having humans help fix the AI's mistakes, they can improve the technology and avoid future mishaps like the one that caused the company's stock price to drop.

Q&As

What is the Google AI tool 'Bard'?
Bard is Google's artificial intelligence search tool.

What do Google executives expect employees to do to help fix bad responses from Bard?
Google executives expect employees to rewrite bad responses from Bard.

How is Google incentivizing employees to test Bard?
Google is incentivizing employees to test Bard by offering them an internal badge and a chance to meet directly with the Bard team.

What are the do's and don'ts that Google provides in the document to employees?
The do's and don'ts that Google provides in the document to employees include keeping responses polite, casual and approachable, avoiding making presumptions based on race, nationality, gender, age, religion, sexual orientation, political ideology, location, or similar categories, and not describing Bard as a person, implying emotion, or claiming to have human-like experiences.

How did Google's stock price react after the announcement of their new AI technology?
Google's stock price dropped nearly 9% after the announcement of their new AI technology.

AI Comments

👍 It is great that Google is leaning on the knowledge of humans to help clean up the AI's mistakes. It is also encouraging that Google is incentivizing employees to test and provide feedback on the AI with a badge and a chance to meet with the Bard team.

👎 Google's rollout of the AI was rushed and botched, leading to a drop in the stock price. Additionally, the instructions to employees on how to fix bad responses can be too restrictive as they are told to keep responses "neutral" and not "imply emotion."

AI Discussion

Me: It's about Google asking employees to rewrite bad responses from their AI tool, Bard. They're doing it to try and improve the AI's accuracy.

Friend: Interesting. What are the implications of this?

Me: Well, I think it shows that AI technology is still quite flawed and that humans are still necessary in providing input and improving it. It also suggests that AI is still a long way from being able to replace humans in certain tasks. On the other hand, it also presents an opportunity for employees to gain experience and be involved in the development of a cutting-edge technology.

Action items

Research other AI tools and their capabilities to compare to Google's Bard.
Create a list of potential topics and questions to test Bard's responses.
Develop a plan to provide feedback on Bard's responses, including guidelines for rewriting responses and providing neutral, unopinionated answers.

Technical terms

Google: Google is an American multinational technology company specializing in Internet-related services and products.
Bard: Bard is a new artificial intelligence search tool developed by Google.
ChatGPT: ChatGPT is a competitor to Google's Bard artificial intelligence search tool.
Do's and Don'ts: A set of instructions or guidelines that explain what should and should not be done in a particular situation.
Rewrite: To write something again in a different way.
Neutral: Not taking sides in a dispute or argument; impartial.
Moma Badge: An internal badge given to employees who contribute to the testing and improvement of Google's Bard artificial intelligence search tool.
Thumbs Down: A gesture used to express disapproval or dislike.