Apple Execs on AI Setbacks, What Went Wrong with Siri and More (Full Interview) | WSJ

Key Concepts

Apple Intelligence, Siri, Large Language Models (LLMs), Private Cloud Compute (PCC), Semantic Index, Liquid Glass, iOS, iPadOS, macOS, Multimodal Computing, Generative AI, ChatGPT, Xcode, Visual Intelligence, Apple Silicon, User Interface (UI), Customer Satisfaction, Tariffs.

Siri and Apple Intelligence

• Siri's Evolution: Last year's WWDC showcased a smarter, AI-driven Siri powered by Apple Intelligence. However, only a portion (4 minutes out of 8) of the Siri features demonstrated were shipped. The delivered features included a new Siri UI, typing to Siri, improved fluency, better conversational context understanding, and enhanced product knowledge.
• Unshipped Features: Significant features didn't meet Apple's quality standards, leading to a shift to "V2" of their architecture. The initial "V1" architecture, while functional, lacked the required reliability for open-ended queries and personalized knowledge.
• Not Vaporware: The demonstrated Siri was based on real, working software with a large language model and semantic search. The decision to delay was due to unacceptable error rates.
• Marketing Focus: Marketing efforts primarily emphasized Apple Intelligence, encompassing over 20 features, rather than solely focusing on Siri. The Bella Ramsey ad was a specific instance related to Siri capabilities.
• Apple Intelligence Strategy: Apple Intelligence aims to integrate generative AI across the operating system, enhancing existing features without being a standalone "chatbot" app.
• Competition: Apple aims for Siri to be superior to competitors but prioritizes delivering it "the right way," focusing on integration, personalization, and privacy.
• Future Plans: Apple is hesitant to disclose specific timelines or future features to avoid missetting expectations.

AI and Model Usage

• Apple's AI Strategy: Apple's approach to AI involves using generative AI as an enabling technology for features across its operating systems.
• Model Integration: Apple already uses its models for writing tools (e.g., making text concise, creating bulleted lists) in private cloud compute.
• ChatGPT Integration: Apple integrates ChatGPT for specific capabilities like visual intelligence and image generation, accessible within Apple apps and third-party apps.
• Xcode Integration: Developers can use ChatGPT and other models (e.g., Anthropic) for coding in Xcode.
• Model Development: Apple is actively developing its models, with the private cloud compute model being "GPT-4o class" in many regards.
• No Chatbot Focus: Apple's goal is not to create another chatbot but to integrate AI capabilities across its operating system and features.

Liquid Glass

• Material Properties: Glass offers useful properties for user interfaces, especially adaptive glass, allowing content to feel open and controls to be inset.
• Adaptive Qualities: Glass can transmit translucency, adapt its properties for contrast, and enhance the UI's visual appeal.
• Technological Advancements: Recent hardware advancements, including processing power (Apple Silicon) and high-definition HDR displays, enable the implementation of Liquid Glass.
• Design Consistency: Liquid Glass aims to bring a consistent design language across Apple's entire product line.
• VisionOS Inspiration: The design is inspired by the work done on VisionOS.

iPadOS and macOS

• Distinct Products: iPad and Mac are distinct products with different centers: iPad optimized for touch, Mac for keyboard and trackpad.
• Evolving iPad Usage: iPad usage has evolved from consumption to productivity, leading to the adoption of some Mac idioms.
• No Convergence: Apple explicitly states (with emphasis) that there are no plans to merge iPadOS and macOS.

The Future of Computing

• Multimodal Future: Apple envisions a multimodal future where users interact with devices using sight, touch, and speech.
• Integrated Interaction: The ultimate future involves seamless integration of these modalities, allowing users to direct devices naturally.
• Wearable Devices: Apple already has personal wearable devices like the Apple Watch and iPhone that are always with the user.

Tariffs and Pricing

• Monitoring Tariffs: Apple is closely monitoring tariffs and their potential impact on product pricing.
• No Announcements: There are no specific announcements regarding price increases due to tariffs.

Apple's Current Position

• Experience and Perspective: The interviewees, with extensive experience at Apple, acknowledge the ups and downs the company has faced.
• Focus on Great Products: Apple's strategy remains focused on creating great products and communicating their value to customers.
• Strong Business: The business is strong, with high customer satisfaction across product lines.
• Excitement for the Future: There is excitement about upcoming products and the positive reception to recent announcements.

Rapid Fire Questions

• "Slow Horses" or "Severance"? "Severance" is number one, but "Slow Horses" is close.
• iPad or Mac? Both.
• Texting Style? Big paragraph (for one interviewee), text return indicating Mac usage (for the other).

Conclusion

Apple is navigating the evolving landscape of AI and computing by focusing on integrating AI into existing products, prioritizing user experience and privacy, and maintaining a commitment to high-quality hardware and software. While facing challenges like competition and potential tariffs, Apple remains confident in its strategy of creating great products and delivering them to customers.