In December 2023, Apple quietly released model libraries and frameworks under MLX group that will run on its native silicon, potentially bringing generative AI tricks to the Mac lineup in the same way that Qualcomm is trying to serve with its Snapdragon X Elite platform for Windows on Arm machines. Another Bloomberg report notes that Apple has expanded its internal team tasked with testing generative AI features, with an overarching goal of releasing these tools for developers in 2024. Rumors suggest that Apple’s first wave of generative AI features will arrive with iOS 18; scheduled to roll out later this year.
Earlier this year, Apple researchers also detailed a generative AI tool called Keyframer in a paper. This tool essentially allows users to create animation content the way models like Dall-E and Sora make images and videos, respectively. It inherently relies on OpenAI’s GPT-4 model, but adds vector graphics into the mix that work on still images.
Apple’s team also released a model paper that details an AI tool which lets users edit images using plain language description. Interestingly, Qualcomm’s latest flagship Snapdragon also touts a similar feature that enables voice-assisted media editing. In September, The Information (via The Verge) reported that Apple is working on “foundation models” aimed at improving Siri — supposedly in the same way that Gemini is pushing Google Assistant — and that the company was reportedly burning “millions of dollars” each day on AI training.
