CommentComment

Did you catch Google I/O this week? Itā€™s Always Interesting to see what the Android Inventors have been up to during the past Annual Interval. šŸ˜‚ This 17-minute recap will get you up to speed if you missed it, but itā€™s safe to say it had a theme. Mobile features didnā€™t make the big headlines this year but were still present with Android Studio getting AI-powered code assistance, a new on-device Gemini model, and updates to Jetpack Compose.

But why am I talking about that here? Well, with Tim Cook saying they are working on generative AI features thatā€™ll debut ā€œlater this yearā€ and Greg Joswiak dropping huge hints when announcing the conference, itā€™s easy to see how WWDC could be similar to I/O this year.

I thought it might be fun to think through what that could mean, but to keep this short Iā€™ll limit myself to three thoughts:

First, Siri. Itā€™s an obvious target for an upgrade and generative AI is well suited to the task. My only worry is that it will need to be meaningfully ahead of the competition to overcome the bias of negative opinions people have about current generation Siri, and the competition took a huge step forward earlier this week. šŸ˜¬ I still hope this happens, though.

Secondly, will we see a Copilot-esque feature in Xcode? One thing I would bet on is that if there is any AI in Xcode, it will be strictly on-device. I find it hard to believe any large company allows its source code to be transmitted to someone elseā€™s server with every keypress, but I find it beyond belief that Apple would. If something like this ships in Xcode 16, Iā€™d also be curious to know what training data it uses. I still worry a little about GPL training data in Copilot, so I hope itā€™s not based on OpenAIā€™s Codex as Copilot is. I also canā€™t see Apple using internal Swift code as training data, can you? There are lots of difficult questions here, so Iā€™ll be fascinated to get more details if this feature ships.

My final thought is a long shot. Itā€™s not a prediction because I donā€™t think itā€™ll happen. Itā€™s more of a hope, and Iā€™m including it here because I think Apple would do a great job with it. Iā€™d love to see a painting app that uses generative AI to work with your brush strokes to create art. It wouldnā€™t generate images from text but would use AI models to help you improve your sketching and painting skills. There are many ways that an app like this could work and differing levels of assistance it could provide so I wonā€™t speculate on anything more specific, but this is my long-shot hope. This is by far the least likely to happen, mainly because if it existed it would have been announced at last weekā€™s iPad event alongside the new pencil. I donā€™t care, though. Itā€™s on my list!

Letā€™s see how wrong I am in a few weeks! šŸ˜¬

Dave Verwer  

News


Tools

Code



Design

And finally...

Someone proud of a detail they thought of in their forgery would point this out.

Great detective work! šŸ•µļøā€ā™‚ļø