Engineers evaluate and approve the new code, which enables them to accomplish more and move more quickly.
During the third quarter earnings call, Sundar Pichai, the CEO of Google, disclosed that AI is responsible for the generation of over 25% of the company’s new code.
He stated that engineers evaluate and approve the code, which enables them to accomplish more and move more quickly.
“We are also utilizing AI internally to enhance our coding processes, increasing productivity and efficiency,” he stated.
Over the past year, Google has decreased its workforce by over 1,000, compared to the same period in 2023.
According to Pichai, Google’s differentiated complete stack approach to AI innovation, which is currently operating geometry dash at scale, has placed the company in a unique position to lead in the AI era.
He further stated that this encompasses a comprehensive global reach through products and platforms, world-class research teams, and a robust AI infrastructure.
The call also observed that Google’s profits have increased across the board, including cloud, services, and Alphabet revenue. The total revenue has reached $88.27 billion, a 15% increase from the previous year’s $76.69 billion, with search and cloud services being the primary contributors.
“LG AI Research reduced inference processing time for its multimodal model by over 50% and operating costs by 72% by utilizing a combination of our TPUs and GPUs,” Pichai stated.
It was also disclosed that YouTube’s ad and subscription revenue for the previous four quarters exceeded $50 billion for the first time.
Study: Cybercriminals Utilize Generative AI to Compose Malware Code
Pichai also emphasized that Google had substantially reduced machine costs per search query by utilizing its AI Overview application, which offers a snapshot of searches with key information and links that are generated by AI.
He stated that the company has doubled the scale of its custom Gemini model and reduced costs for these queries by over 90% in the 18 months since first testing AI Overviews through hardware, engineering, and technical breakthroughs. The latter is a collection of multimodal large language models that were created by Google DeepMind.
Additionally, AI Overviews has begun to expand to over 100 new countries and territories, and it will now reach over 1 billion users on a monthly basis.
Gemini models are implemented in all seven of Google’s products and platforms, which have a monthly user base exceeding two billion. This encompasses Google Maps, the most recent product to surpass the billion-user threshold.
In response to substantial demand, Pichai announced that Gemini will be accessible on GitHub Copilot in addition to Google’s proprietary platforms.
He also stated that the company is developing experiences in which AI can observe and reason about the world around the user, such as a future product called Project Astra, which is expected to be available in 2025.