Google employees are testing the company’s AI chatbot Bard ahead of release.
Employees also have access to a better version, the “Big Bard”.
Big Bard produces much richer and more human-like responses.
We are currently testing machine translations of articles by our US colleagues at Insider as an added service to our readers. This article has been automatically translated and reviewed by an editor. We welcome feedback at the end of the article.
Google has asked its employees to test and improve internal chatbot Bard before releasing it to the public. Many employees are already playing with a better version.
Google employees built “Big Bard,” as the company calls it internally, based on the same language model as Bard, known as LaMDA. The difference: Big Bard has larger parameters that make him seem smarter.
Business Insider looked at examples of users asking similar questions to both versions. Big Bard, however, provided richer and more human-like answers. Also, it is generally more talkative, informal, and prone to swearing.
When Google unveiled the AI Bard, Google CEO Sundar Pichai said the first version released to the public will be based on a “lightweight model” from LaMDA. This requires less computing power. Big Bard appears to be a preview of what a more advanced version of the chatbot might look like.
What should be brought to market?
Google may want to release a limited version of Bard for cost reasons. John Hennessy, the chairman of Google’s parent company Alphabet, recently said that an AI search could potentially cost Google 10 times more than a regular keyword search.
Google is also wary of public glitches that could create negative headlines. This includes, for example, inaccurate answers or dubious behavior, which could be prevented by a limited start.
When employees test Big Bard, they receive a notification directing them to the regular version of Bard. Some executives have asked their employees to only work with the standard model, according to an employee. An earlier version, tested internally, was called the Apprentice Bard.
The race against AI doesn’t seem to be slowing down. Andreas Braun, the CTO of Microsoft Germany, just announced that OpenAI will release the next version of its language model, GPT-4, next week. GPT-4 will be multimodal, meaning it will also include video, images, and audio. A Google spokesperson did not respond to a request for comment.
Did you like the translation?We look forward to your feedback by email [email protected]!
Read the original article Insider.