A scene from the movie Top Gun showing the character Goose. CBS Photo Archive/Getty Images
Google has a large internal language model called “Goose” that is designed to make employees more productive.
Goose is trained based on “25 years of technical experience at Google†.
The company is using AI as part of a larger efficiency offensive.
As Google builds artificial intelligence into all sorts of products, it is also relying on AI to make its business more efficient. The company has introduced a large language model called Goose that is intended only for employees. It will help develop new products, according to internal documents seen by Business Insider.
In the documents, Goose is described as a “descendant of Gemini,” the company’s major new language model. Goose is “trained on the sum of 25 years of technical experience at Google,” one of the documents states.
“It can answer questions about Google-specific technologies, write code using internal tech stacks, and supports novel features such as editing code based on natural language prompts,” says an internal memo about Goose.
Read too
This woman landed a job at Google with the help of her viral application video
Google is trying to become more efficient
Internal use of AI could be a key factor in Google’s new efficiency efforts, which have resulted in thousands of job cuts and a series of team shakeups over the past 13 months. Google’s chief financial officer, Ruth Porat, hinted at such initiatives during the company’s most recent earnings release. She told analysts that Google is working to “slow cost growth through a number of measures,” including “streamlining operations at Alphabet through the use of AI.”
An internal document says Goose is part of a plan to “bring AI to every stage of the product development process.” A Google spokesman did not respond to a request for comment.
Goose seems to already be available to at least some employees. However, it is not clear whether it is fully functional in terms of its capabilities. One document states that Goose “will also be the first general-purpose LLM approved for internal coding at Google.”
“Goose has a 28k token context window, which makes it particularly valuable for development tasks,” it continues. Token context windows refer to the amount of text, numbers, and other information that a language model can consider when processing a query. This is especially important for coding, which can involve hundreds of thousands of lines of code.
The documents note that the development of Goose was a joint project between Google Brain, DeepMind and Google’s internal infrastructure teams. The company merged Brain and DeepMind last year to create Gemini, which in turn named after the two teams that developed it.
Read too
The Google CEO reads this unknown news website every day – Zuckerberg and Nadella also read it often
AI isn’t taking away jobs – yet
Google isn’t the only tech company using its own AI models and products to boost productivity. Microsoft recently launched its Copilot tool for internal teams, Business Insider reports.
And Goose isn’t the only example of how Google is using AI to become more efficient. For example, the company uses AI to reduce the energy required to cool its data centers. It’s also pushing more and more advertisers to use automated AI products to decide where to spend their money across Google products.
At least for now, Google insists that AI won’t take away employees’ jobs. “We’re not restructuring because AI is taking jobs,” Google Chief Business Officer Philipp Schindler said after Business Insider reported layoffs from Google’s ad sales team as part of a restructuring.
Goose could become increasingly useful to employees. The tool was developed with a “focus on new engineering capabilities,” one of the documents says, suggesting it could go far beyond the existing development tools offered to employees.
And if Googlers have specific development questions while using Goose, they are encouraged to reach out to the company’s internal chatbot called Duckie.
Read the original article in English here.