The note, recently seen by MacRumors, also suggests that Apple is planning to build several hundred AI servers this year, with plans for even more the following year. If reports are to be believed, Apple has plans to bring generative AI to its hardware by late 2024. The servers will use a combination of cloud-based AI and “edge AI,” which processes data on-device.
The Information earlier reported that the company is investing millions of dollars daily to train its large-language models (LLM), which will be incorporated by the Siri team to make complex shortcut integrations more accessible. Although most of this investment will focus on AppleCare customers, the language models will be available for wider use.
Apple has a generative AI tool internally, called “AppleGPT,” built on its own large language model called “Ajax.” The framework is based on Google JAX, which is designed to accelerate research in machine learning. AppleGPT can summarise the text and answer questions based on data it has been fed, similar to other chatbots like Bard, ChatGPT, and Bing AI.
The iPhone maker has been actively hiring engineers and scientists with expertise in artificial intelligence, particularly in generative AI. Senior Vice President of Software Engineering, Craig Federighi and John Giannandrea, Senior Vice President of Machine Learning and AI, are leading this effort.
Although Apple’s plans for the chatbot are not yet clear, Bloomberg reported that the company may make a “significant AI-related announcement” sometime next year.
Pu notes that Apple has been taking a slow approach as it figures out a way to implement generative AI in a way that aligns with its principles around user security and privacy. In an investor call earlier this year, CEO Tim Cook acknowledged the potential of generative AI while also recognising the need to address its concerns. Cook has also mentioned the importance of self-regulation in the tech industry.