Published on October 24, 2023, 12:33 am
The growing popularity of generative AI has taken the big technology players by surprise. From the widely known ChatGPT to AI artwork like Dall-E, Stable Diffusion, and Midjourney, generative AI has captured the attention of both users and tech giants alike. It even made its way into popular culture with a dedicated episode on South Park. Notable figures like Tom Hanks have also expressed concern over AI-generated fakes.
Companies like Microsoft, Amazon, Google, and Apple are racing to incorporate generative AI into their products and services. Apple, in particular, has been working behind the scenes to add significant AI features to all its major offerings. The integration of AI is not new for Apple as it already utilizes machine learning and artificial intelligence in various aspects such as photography, Siri, crash detection, facial recognition, sleep tracking, and more.
However, most of these existing features do not fall under the category of generative AI. With the release of iOS 17, Apple introduced a new transformer model that significantly improved dictation and speech recognition in Siri. This upgrade showcased Apple’s first use of transformer models, which are foundational for generative AI applications.
According to a report by Mark Gurman for Bloomberg’s Power On newsletter, Apple executives were initially caught off guard by the sudden surge in interest surrounding generative AI. As a result, they have been playing catch-up since late last year. Other tech giants like Microsoft and Google responded quickly by partnering with OpenAI and pushing out their own AI technologies respectively. In comparison, Apple focused on improving autocorrect functionality.
Gurman’s report suggests that Apple now has an extensive endeavor underway to bring generative AI features alongside other advancements to its releases in 2024. This means that users can expect improved AI capabilities in future iterations of iOS (iOS 18) and macOS (macOS 15), as well as across various Apple apps and services.
Apple has been working with its own large language model called Ajax for over a year, and it has developed an internal chatbot similar to ChatGPT, referred to as “AppleGPT.” While Apple may not directly release a ChatGPT competitor, the Ajax technology could find application in other products. Additionally, Apple is exploring various generative AI features under scrutiny.
Senior VP of AI at Apple, John Giannandrea, along with Craig Federighi (Senior VP of software engineering), is spearheading the effort to bring AI to as many areas as possible by 2024. Apple has allocated around $1 billion in research and development for its AI initiatives over the next year.
The implications for users are yet to be fully revealed, but Giannandrea and his team are expected to deliver a new version of Siri that seamlessly integrates Apple’s advanced AI technology. Gurman notes that iOS 18 will likely be filled with features powered by Apple’s large language model. Some anticipated features include improved suggested replies in the Messages app, an updated version of Xcode that assists developers with generating code automatically (similar to Github’s Copilot feature), an AI auto-generated playlist maker or DJ for Apple Music, and generative AI tools integrated into the iWork suite (Pages, Numbers, Keynote) akin to what Microsoft is implementing in Word and PowerPoint.
The possibilities for generative AI are limitless. Adobe has already implemented it into Photoshop, facilitating tasks such as expanding photos beyond their original framing or removing or adding objects. Similar applications could enhance users’ photo experiences on iPhones. Generative AI might also offer background audio tracks for iMovie and Final Cut or even royalty-free intro/outro music for podcasts created in GarageBand.
Regarding the implementation of certain features, there is internal discussion within Apple about whether they should be processed on-device or on Apple’s cloud servers. On-device processing prioritizes privacy while more complex AI models require robust server hardware. On-device processing, however, poses challenges in terms of updates, training, and deployment compared to the server-based approach. Gurman predicts that Apple will adopt a mix of both approaches, with certain features processed on-device and others in the cloud.
It’s safe to say that Apple’s push towards generative AI holds exciting prospects for users. As this technology continues to evolve, we can expect Apple to deliver innovative experiences that harness the power of AI while ensuring user privacy and convenience.