Apple’s research labs have quietly been working on an AI-driven chatbot project, internally referred to as “Apple GPT,” according to recent findings from Bloomberg. While the tech giant is known for its privacy-centric stance, the intended usage and public release of this technology remain obscure.
The chatbot is powered by Apple’s in-house large language model (LLM) system, referred to as “Ajax,” that utilizes Google Cloud infrastructure and Google’s JAX, a renowned framework designed for speeding up machine learning research. Insiders reveal that multiple teams at Apple are collaborating on this project, tackling not just the technicalities but also potential privacy complications.
Despite its slow progress, Apple is not to be counted out in the generative AI race, a domain where tech behemoths like Meta, Microsoft, and Google have been making strides, launching generative AI solutions for businesses and consumers alike. While Apple has refrained from endorsing the use of ChatGPT among its employees, the Ajax-fueled chatbot has reportedly been in use internally, serving as a testbed for integrating machine learning more efficiently.
Despite the conspicuous silence, Apple has been integrating AI into its software for years, with Siri, the voice assistant, serving as the star player. Siri, which trailblazed the voice assistant trend, has been critiqued for its comparative limitations, prompting Apple to ramp up its efforts. In a strategic move, Apple roped in John Giannandrea, the former AI and search lead at Google, in 2018 to superintend Siri’s development and the broader machine learning teams. Bloomberg reports that Giannandrea, in tandem with Apple’s SVP of Software Engineering, Craig Federighi, is at the helm of Apple’s AI push.
Apple CEO, Tim Cook, recently addressed the subject of AI technology in a Good Morning America interview, describing it as an area of “intense focus” for the company. He also voiced concerns about AI products during an earnings call, underscoring the need to resolve “multiple issues.” Interestingly, other corporations have started leveraging different LLMs for the benefit of startups and researchers. For instance, Meta announced the accessibility of its LLM, LLaMA 2, on Microsoft’s Azure platform, while Microsoft itself has integrated OpenAI’s GPT model into its Bing search product.
Though Apple’s trajectory in the AI domain may seem hazy, Bloomberg’s insiders suggest that the company is gearing up to make a “major AI-related announcement” in 2024. It remains to be seen whether this implies the launch of their new chatbot or a broader AI strategy. Yet, this news underscores Apple’s determination to make significant strides in the evolving world of AI, while ensuring the company’s core values of privacy and user-centric design are upheld.