Apple GPT may come sooner than expected. Apple's technique for bringing large language models (LLM) to iPhone and other devices has been revealed.
Apple artificial intelligence (AI) researchers said that with an innovative technique for using flash memory, it will be possible to use large language models (LLMs) on iPhones and other Apple devices with limited memory. In this context, Apple GPT may come to Apple devices sooner than expected.
LLM models are coming to iPhone
LLM (Large Language Model) - based chatbots like ChatGPT and Claude require an incredible amount of data and memory. Chatbots typically require huge amounts of memory to run, which is a problem for devices with limited memory capacity, such as iPhones. This problem could be solved if the AI model uses flash memory (the same memory that holds apps and photos) to store data.
A new research paper shows that flash storage on mobile devices is more efficient than conventionally used RAM for training large language models. The method found is to remove limitations by using two basic techniques that minimise data transfer and maximise flash memory throughput: Windowing and Row-Column Bundling. In the first, the AI model reuses some of the data it has already processed instead of loading new data each time. This reduces the need for constant memory retrieval, making the process faster and smoother. In the second, the data is grouped more efficiently, allowing it to read from flash memory faster, which means the AI can understand and create language faster.
iOS 18 will be an artificial intelligence update
The leap in AI efficiency will open up new possibilities for new iPhones, including improved Siri capabilities, real-time language translation, and advanced AI-powered features in photography and augmented reality. The technology also paves the way for iPhones to run complex AI assistants and chatbots on the device. Apple's work on generative AI will eventually be incorporated into Siri. Apple is also known to be working on a smarter version of Siri that is deeply integrated with AI.
Apple is reportedly developing its own generative artificial intelligence model called "Ajax". Designed as a rival to OpenAI's GPT-3 and GPT-4, Ajax works on 200 billion parameters. It is claimed that iPhone and iPad will have some kind of generative artificial intelligence feature when iOS 18 is released.
