Artificial intelligence has become one of those emerging technologies that everyone wants to dip their hands into. While many adopted this trend as soon as it reached the mainstream, some were still on the fence about it. But it appears even they are starting to convert. Recently, Twitter chief Elon Musk, one of the most vocal critics of AI, started his own startup in the sector called xAI. And now, Apple, a company that has avoided even using the word AI, is working on its own foundational model that has reportedly already deployed internally a generative AI platform called Apple GPT.
The information comes from Bloomberg’s Mark Gurman, a long-time Apple tipster, who said in a tweet, “Apple races to build next-generation, Generative AI features to catch up with OpenAI and Google. The company has developed a framework called “Ajax” to build LLMs and has deployed “Apple GPT” internally. It’s aiming for a consumer product next year”.
Ajax appears to be an in-house foundational model that can help build large language models, one of them being Apple GPT, as per the report. It is not known whether the name Apple GPT is just being used for internal purposes, or it can also be the final name once the consumer product comes out.
Apple’s AI pursuit can be complicated
The problem lies within data sourcing to create generative AI, and possibly a chatbot. Every other company that has come up with its own generative AI, be it Microsoft, Google, OpenAI, or recently Meta, all rely on an integrated cloud-based storage, that the company has access to. This data is massive and can be used to train AI models. However, Apple does not function this way.
A report by TechRadar explains that Apple takes the machine learning approach, where the Apple silicon locally learns about the device owner and that’s all the data it gets. In fact, Apple has a very heavy emphasis on data privacy and data security and has multiple times highlighted how it does not keep access to user data. Even at the launch of the Apple Watch Series 8 last year, Apple mentioned that none of the health data will be stored on Apple servers and only the user will have access to it.
A system built so robustly around the idea of not stockpiling data goes against the requirements needed to build an AI platform.
The TechRadar report highlights that for this reason, it is unlikely that Apple GPT will ever see the light of the day and is likely just an internal project to further develop its technology.
Either way, we are excited to see how the development of AI turns out for Apple, and whether we ever get to see an integrated Apple GPT on iPhones.