Apple’s Late Arrival; Meta Outlines Ways to Make Money from AI, Including Charging Users

If you weren’t careful, you may have missed some important news from Apple that was overshadowed by Meta Platforms’ earnings yesterday. 

The iPhone-maker released a new family of open-source language models, dubbed OpenELM, on Hugging Face. The models, ranging in size from 270 million to 3 billion parameters (as a reminder, “parameters’ refers to the “settings” that determine how AI models respond to questions) are significantly smaller than open-source releases from other big tech companies like Meta so far.

That gives us a window into Apple’s most likely strategy: developing artificial intelligence software that’s small and light-weight enough to run on its devices. You can imagine a future where a “personal large language model” running on your iPhone can remember your unique preferences and habits and can help you complete tasks like ordering a Starbucks coffee for you each morning, or drafting out messages to friends and colleagues in formal or informal language depending on the recipient.