WWDC: Intelligence on Apple Intelligence

Apple will show us what it has achieved in artificial intelligence in a few days at its Worldwide Developers Conference (WWDC), which begins with a June 8 keynote speech.

As is customary at this time of year, rumors have emerged to prepare the market for what Apple has planned.

Privacy as Apple’s USP

Some of these reports suggest Apple will be making a big push to champion privacy in its execution of AI. It will stress that the capacity to run AI services on device is a huge advantage, as while cloud-based processing will still be necessary for some tasks, it will remain possible to process some AI models entirely on the device. 

Apple has been working with Google to develop new AI models and is thought to be on the hunt for strategic acquisitions to help broaden its AI offerings. 

Reporting from the Information tells us Apple has also agreed to permit some AI requests to use Nvidia’s confidential compute tech in Google Cloud, in addition to its own Private Cloud Compute infrastructure.

Ultimately, Apple plans to make privacy and edge AI key planks in its approach to AI at WWDC, it seems.

How the new Siri will work

Beyond strategy, details concerning how these new features will work is also sneaking through. Bloomberg has created a set of mocked-up screenshots to show what the reporter thinks the new Siri app will be like in iOS 27, including how a new Search or Ask functionality, basically an upgraded version of Siri Suggestions, works.

The mock-ups depict a new Siri interaction on screen, which includes a choice to use Apple Intelligence or an installed and supported third-party AI model to fulfil a search written in the search bar or up on the Dynamic Island. A new Extensions feature will let you select and deploy additional services to replace ChatGPT.

To access the features, you’ll be able to say Siri, or hold down the power button, but you’ll also be able to swipe down from the top centre of the iPhone screen to invoke the service.

Siri will of course gain contextual understanding, which means it will be aware of and capable of working with what you have on-screen. That personalization extends to things like scheduling appointments, writing emails, or texts.

There are other features:

• Siri will support conversational AI, so you will be able to refine and ask questions as you search.
• There will be a messaging history, which can be set to expire.
• To a great extent, the interface will feel familiar to anyone who has used an AI chatbot.
• You add images and other files using a paperclip option that sits to the side of the search bar on an iPhone.

What about the developers?

For developers, Apple is likely to extend the number of AI services it make available via the Foundation Models framework. This will empower developers to weave complex AI features within their apps, enabling them to build AI tools for use on the edge. 

We also anticipate a Visual Intelligence API which will let developers build apps that can get information about anything you point your device at, while personal context APIs, which I think may be limited at first, suggest developers may be able to achieve powerful new personalization within their apps. 

Just last week, Apple revealed a host of Accessibility tools based on AI it will introduce later this year, which included a powerful new tool  called Image Explorer in VoiceOver lets you use Apple Intelligence to generate detailed descriptions of images held across your system, and those images can be documents, bills, receipts.

You can follow me on social media! Join me on BlueSky,  LinkedIn, and Mastodon.