OpenELM is a new open source AI framework from Apple made for on-device processing.
Two things make this move by Apple interesting:
- Open Source
- On-Device
To the first point, Apple doesn’t usually release open source code. Usually, Apple is a prime example of a “walled garden” and “closed ecosystem”. To me, this open source release can be interpreted as a plea for help from Apple for developers globally to help them catch up with AI development. This is further supported by the recent conversations Apple has held with Google to license their AI (Gemini).
To the second point, Apple hopes the privacy benefits for on-device AI prove to be compelling. The alternative would be AI that relies on a large data center for processing. Large data centers are the wheelhouse of Google, Microsoft, and Amazon; however, Apple’s data center size doesn’t make the top 10. Therefore, Apple hopes to champion the privacy benefits of on-device AI. I’ve posted before about how well AI will get to know us; however, AI using large data centers will likely have much higher capabilities compared to the limitations of on-device AI models.
My point of view on what this means for the future is that we will see Apple’s AI capabilities in the September/October iOS update fall far short of AI offerings from Google and Microsoft.