April 19, 2024
Apple Prefers On-Device AI Processing
AI

Apple Prefers On-Device AI Processing

Apple’s recent research on implementing large language models (LLMs) on smartphones suggests a strategic move to catch up with Silicon Valley competitors in generative artificial intelligence (AI). The paper, titled “LLM in a Flash,” addresses a “computational bottleneck,” paving the way for effective LLM inference on devices with limited memory. This approach contrasts with the typical operation of chatbots and LLMs in data centers with more computing power than an iPhone.

Published on December 12 and highlighted by Hugging Face, the paper is Apple’s second on generative AI this month, following efforts to enable models like Stable Diffusion on its custom chips. The push for new AI features aims to revitalize the smartphone market, which experienced a 5% decline in shipments, according to Counterpoint Research.

Apple, often perceived as lagging in generative AI compared to rivals, indicates a focus on AI running directly on iPhones. This diverges from approaches by Microsoft and Google, emphasizing cloud-based generative AI services. While competitors plan AI-focused smartphones, Apple’s research signals a potential shift towards on-device AI.

Qualcomm CEO Cristiano Amon anticipates a new smartphone experience with generative AI use cases, expecting launches in early 2024. The rise of AI-focused smartphones is estimated to exceed 100 million shipments in 2024, with 40% of new devices offering such capabilities by 2027, according to Counterpoint.

Running large AI models on personal devices poses technical challenges due to limited computing resources, but Apple’s approach could enhance AI assistant responsiveness and privacy. Google has also introduced a smartphone-native LLM. Apple’s experiment, optimizing LLM inference on personal devices, holds potential for future device and application advancements as LLMs grow in size and complexity.

Photo by Kartikey Das

Related posts

Government Action Urged to Combat AI Deepfakes: Andrew Yang Leads Global Call

Christian Green

New AI Safety Standards: Biden’s Executive Order Prioritizes Privacy and National Security

Harper Hall

G7 Summit: Agreement on AI Code to Ensure Safe and Trustworthy Development

Bran Lopez

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More