About this video
Apple is finally admitting that Siri has been a legacy of failure. In this video, I dive deep into the WWDC keynote to see if iOS 27 and the new Siri AI can actually deliver on the promises made years ago. We cover everything from the new onscreen awareness and advanced photo reframing to the proactive writing tools and the new dedicated Siri chatbot app. Key Takeaways: - Siri AI now focuses on personal context and natural language actions across all Apple devices. - Onscreen awareness allows Siri to interact with exactly what you are viewing in real time. - New photo features include local model reframing and server side image generation. - Natural language can now be used to build complex structures in the Shortcuts app. - Advanced AI features are limited to iPhone 17 Pro, M3+ Macs, and M4+ iPads.
Apple WWDC Deep Dive: Is Siri AI the Real Deal?
Apple is finally admitting that Siri has been a legacy of failure for the better part of a decade. During this week's WWDC keynote, the tech giant unveiled its latest attempt to reclaim the throne of digital assistants with the introduction of Siri AI and iOS 27. But after the disastrous attempts at 'personal context' back in 2024, is there any reason to believe they have got it right this time?
The Siri Evolution
The standout announcement was the categorisation of Siri AI as a system that encompasses personal context and natural language actions. Unlike previous iterations, this version aims to understand the files, interactions, and applications across your Mac and iPhone to create a cohesive digital experience. For those with the hardware to support it, specifically the iPhone 17 Pro and M-series chips, the promise of onscreen awareness is a game changer. Being able to ask Siri to action something based on what you are looking at in the moment could finally make the assistant feel like a proactive partner rather than a voice activated timer.
Visual Intelligence and Photo Magic
Apple also showcased significant enhancements to photo editing and visual recognition. By leveraging local models for reframing and server side models for final rendering, users can now adjust perspective and zoom after a photo has been taken. Furthermore, the expansion of visual intelligence across MacOS and iOS means you can query any screenshot for recipes, products, or information. This is particularly potent on the Vision Pro, where simply looking at an object allows you to ask Siri for physical dimensions or compatibility.
The Privacy Trade-Off
The crux of Apple's strategy remains privacy. With their 21 billion parameter foundation model, Apple is betting that users will choose a slightly less capable AI if it means their data stays on device or in a 'Private Cloud'. While competitors like Google and OpenAI have pushed the boundaries of what agents can do by consuming vast amounts of personal data, Apple is attempting a more tasteful, restrained approach.
Is it Enough?
The missing piece of the puzzle remains true agentic capability. We still haven't seen Siri move seamlessly between third party apps, such as taking a document from an email and autonomously messaging it to a contact on WhatsApp. Until Apple breaks down these silos, Siri AI may just be a very polished version of what we already have. Only time will tell if iOS 27 is the revolution we were promised or just another incremental update in a very expensive ecosystem.
Transcript▾
Apple's WWDC conference is happening this week and they kicked us off with their keynote presentation. This year focuses mostly on performance and there's a lot of safeguard features for parents and their children. In this video, we're going to go over the AI specific stuff that I find interesting or that you might have missed.
They introduced a new offshoot called Siri AI, which encompasses personal context and actions taken on your phone and Mac using natural language. Apple also revealed they're working with Google and Nvidia on their cloud models. Siri AI is available on many products, but the more advanced features require the iPhone 17 Pro, iPhone Air, M3+ Macs, or M4+ iPads.
One of the biggest enhancements is general onscreen awareness. You can trigger Siri at any point and ask it to action things based on what is on your screen. Siri is now much better at maintaining a conversation and following up on questions. They also showed off photo enhancements like reframing using local models and server side rendering for the final image.
The text generation features now include active improvements to writing style and proactive proofreading. Siri visual intelligence is expanding, allowing you to query images or screenshots for recipes and products. On Vision Pro, you can even ask about physical objects in your environment.
A dedicated Siri app now allows for a text based chatbot experience using Apple models. Voice dictation is also seeing improvements, which I'll be comparing to Whisper Flow. In messages and calendars, Siri now understands context better to suggest photos or dates.
Shortcuts can now be created using natural language, making the app much more accessible. Finally, the passwords app can now automatically change compromised passwords across various websites. On MacOS, Siri can index your files for better context, though it only indexes the first few characters. Safari has also gained tab grouping and price drop notifications.
The big question is whether this is the Apple Intelligence we were promised in 2024 or another failed attempt. With a 21 billion parameter model, we will see if the balance of privacy and performance actually works for users.