Maybe so we can talk/chat with our phones themselves when we're bored and lonely? Generate random images tojerkfulfill our entertainment itch? Yeah, I have no clue on why we need an entire neural network on our phones.
Google is planning LLM to Google assistant, to turn it into a real chatbot. For image generation, there are a ton of apps and filters that can leverage this. Yhea that seems shallow, but those kinds of stuff make a killing.On device image generation gen AI, wha, why? What is the point? Most other updates seem, at the very least, neat (If a bit overhyped, I doubt that gaming will actually be "Hyper-realistic"), but Image generation? What's the need, what's the point? Who are they targeting with this?
It seems that on device processing is also more reactive than cloud processing. The latter is favoured when you need to process a massive amount of data or need higher accuracy.