Apple has officially entered the generative artificial intelligence race with the announcement of Apple Intelligence, a suite of AI features deeply integrated into its upcoming operating systems: iOS 18, iPadOS 18, and macOS Sequoia. Revealed during the company’s Worldwide Developers Conference (WWDC) 2024, Apple Intelligence represents a significant strategic move, aiming to blend powerful AI capabilities seamlessly into the user experience while upholding the company’s long-standing commitment to privacy.
Understanding Apple Intelligence
Unlike standalone AI applications, Apple Intelligence is envisioned as a “personal intelligence system.” It leverages the power of large language models (LLMs) and diffusion models directly within the operating system core. This integration allows AI features to understand user context derived from personal data like emails, messages, calendars, photos, and on-screen activity, without necessarily sending that data to the cloud. The system is designed to be helpful and relevant by understanding personal context while prioritizing user privacy.
Apple emphasizes that its approach is different from competitors. It aims not just to provide generative AI tools but to enhance existing applications and workflows, making tasks simpler and more efficient across the Apple ecosystem. The system intelligently determines whether a task can be handled by the relatively smaller, efficient on-device models or requires the power of larger server-based models via its Private Cloud Compute infrastructure or external partners like OpenAI.
Core Features and Capabilities
Apple Intelligence introduces a range of new capabilities across various applications.
Enhanced Writing Tools
System-wide Writing Tools offer features like Rewrite, Proofread, and Summarize. Users can select text in apps like Mail, Notes, Pages, or even third-party apps and ask the AI to adjust the tone (e.g., make it more professional or friendly), check grammar and style, or condense long passages into key points. This aims to streamline communication and content creation directly within the user’s workflow.
Image Playground and Genmoji
For visual creativity, Image Playground allows users to generate images in Animation, Illustration, or Sketch styles based on text descriptions. This feature is integrated into apps like Messages and Notes, and a standalone app will also be available. It allows for rapid visualization and creative expression. Relatedly, Genmoji enables users to create unique emoji characters based on descriptions or even photos of people, offering a new layer of personalized communication.
Siri’s Transformation
Siri receives a substantial upgrade, becoming more conversational, context-aware, and capable. Powered by Apple Intelligence, the new Siri can understand natural language better, maintain context across multiple requests, and take actions within and across apps. Users can type requests to Siri in addition to using voice commands. It gains on-screen awareness, allowing users to ask questions or issue commands related to what’s currently displayed. For instance, a user could ask Siri to “add this address to John’s contact card” when viewing an address in a message.
Mail, Photos, and Notes Enhancements
In Mail, Priority Messages surface the most urgent emails, while Smart Reply suggests quick responses, and long threads can be summarized. Photos benefits from improved search capabilities using natural language (e.g., “find photos of Sarah kayaking at the lake”) and a new Clean Up tool to remove unwanted objects from backgrounds. Notes can now generate summaries of lengthy notes and transcribe and summarize audio recordings.
Privacy and Security Approach
Privacy remains a cornerstone of Apple’s strategy. Many Apple Intelligence features run entirely on-device, leveraging the power of Apple Silicon’s Neural Engine. For more complex requests that require larger models, Apple introduces Private Cloud Compute (PCC). PCC sends only the necessary data to run the task on secure Apple silicon servers, and Apple states this data is never stored or made accessible to Apple itself. The company emphasizes that independent experts can inspect the code running on these servers to verify privacy claims.
Integration with ChatGPT
Recognizing the capabilities of state-of-the-art external models, Apple has partnered with OpenAI to integrate ChatGPT (specifically GPT-4o) into the user experience. When Siri or other tools determine that a request might benefit from ChatGPT’s broader world knowledge or advanced creative capabilities, the system will ask the user for permission before sending the query to ChatGPT. The user’s IP address is obscured, and OpenAI commits not to store the requests. This integration is presented as an opt-in feature, allowing users access to advanced AI without needing a separate account, though subscribers can link their accounts for premium features.
Availability and Device Requirements
Apple Intelligence features will begin rolling out in beta form with the public releases of iOS 18, iPadOS 18, and macOS Sequoia in the fall of 2024, initially available in US English. Broader language support and feature availability will follow over the next year.
Due to the computational demands, particularly for on-device processing, Apple Intelligence will require devices equipped with an A17 Pro chip (iPhone 15 Pro and Pro Max) or an M-series chip (M1, M2, M3, M4) for iPads and Macs. This limitation means older devices will not gain access to these integrated AI features.
Potential Impact
Apple’s entry significantly alters the AI landscape. By integrating AI deeply into its vast ecosystem of devices and software, Apple could rapidly normalize the use of personal AI assistants for hundreds of millions of users. Its focus on privacy through on-device processing and Private Cloud Compute sets a high bar for competitors. The seamless integration aims to make AI less of a distinct tool and more of an invisible enhancement to everyday digital tasks. However, the device requirements create a new hardware upgrade incentive, and the effectiveness and true privacy of the Private Cloud Compute system will face scrutiny. The partnership with OpenAI also signals a pragmatic approach, acknowledging the strengths of external models while maintaining control over the core user experience.