A recent chat on The Circuit podcast gave us a glimpse into how Apple is tackling artificial intelligence with its focus on silicon development. Apple execs Tom Boger and Tim Millet chatted about the company’s path in AI and chip design, sharing some key decisions that helped shape Apple’s current AI strengths.
In 2017, Apple introduced its first Neural Engine, designed to enhance computational photography. However, the company quickly recognized the broader potential of this technology for AI applications.
The same year, following the publication of the influential “Attention is All You Need” paper, which introduced transformer networks, Apple’s silicon team began re-architecting the Neural Engine. This quick response to emerging AI trends proved crucial for future developments.
By 2020, when Apple released the M1 chip as part of its transition to Apple Silicon, the redesigned Neural Engine was capable of running advanced AI networks and hence, this foresight allowed Apple to commit to running AI features across all Macs with Apple Silicon.
The early adaptations in chip design have directly contributed to the current capabilities of Apple Intelligence. This suite of AI features can now run efficiently on devices powered by Apple Silicon, showcasing the company’s commitment to on-device AI processing.
Apple’s approach to AI has been more cautious compared to some of its rivals, but this interview hints that the company has been building the groundwork for AI integration for years.
More here by 9to5Mac.