Apple's AI chatbot project shows promising progress, despite facing critical hallucination issues. Internal testing reveals capabilities comparable to leading competitors like ChatGPT. Under Robby Walker's leadership, the Knowledge project aims to transform Siri into a conversational assistant with internet search capabilities.
Apple has been quietly developing its answer to ChatGPT as part of a strategic push to close the gap in the artificial intelligence race. The Cupertino giant's internal project, nicknamed “Knowledge,” represents a significant evolution for Siri, transforming it from a basic voice assistant into a sophisticated conversational AI. Current testing shows impressive results, though persistent hallucination problems remain a significant obstacle to public release.
The architecture behind Apple's AI chatbot ambitions
Apple's approach to creating a ChatGPT competitor involves developing a comprehensive range of AI models with varying capabilities. The company has engineered models featuring different parameter counts to handle various tasks efficiently.
The current lineup includes models with:
- 3 billion parameters (for basic tasks)
- 7 billion parameters (for intermediate functions)
- 33 billion parameters (for complex reasoning)
- 150 billion parameters (for advanced applications)
The most powerful 150-billion-parameter model operates in the cloud, offering significantly enhanced intelligence compared to the local models powering Apple Intelligence on iPhones. This cloud-based approach allows for more sophisticated processing while maintaining Apple's commitment to privacy.
Former Siri team senior director Robby Walker now leads this ambitious project, bringing valuable expertise to the development process. The Knowledge chatbot aims to differentiate itself by seamlessly integrating with Apple's ecosystem while providing ChatGPT-like information retrieval and assistance capabilities.
Hallucination problems are delaying the public release
Despite considerable progress, Apple's chatbot development faces a critical challenge that has slowed its public debut. The AI system frequently produces hallucinations—confidently presenting fabricated information as fact—a problem that plagued early versions of competing AI systems.
Internal benchmarks reveal that Knowledge's performance metrics approach those of recent ChatGPT iterations in many areas. However, the hallucination issue represents a significant quality control concern for Apple executives, including former Siri head John Giannandrea, who advocates for a cautious approach.
This careful stance reflects Apple's corporate philosophy of releasing polished products rather than rushing features to market. The company appears determined to avoid repeating past mistakes with premature announcements, as with certain Apple Intelligence features.
| Challenge | Impact | Apple's Approach |
| AI Hallucinations | Reduced reliability and accuracy | Delaying the release until it is resolved |
| Competition from established AI | Market pressure to release quickly | Prioritizing quality over speed |
| User expectations | High standards for Apple products | Comprehensive internal testing |
Transforming Siri with advanced vocal capabilities
Beyond the chatbot functionality, Apple is working on enhancing Siri's vocal interactions. The upcoming vocal version of Siri, expected to roll out in fall 2025, aims to deliver a more natural, dynamic conversational experience similar to ChatGPT's voice mode.
This vocal upgrade represents a significant step in Apple's AI strategy, potentially transforming how users interact with their devices. The enhanced Siri would allow for more fluid conversations, moving beyond the rigid command-response pattern that has characterized voice assistants for years.
The integration of these advanced models would enable Siri to:
- Understand complex, multi-part questions
- Maintain context throughout extended conversations
- Respond with more natural, human-like speech patterns
- Access and synthesize information from multiple sources
- Provide more personalized assistance based on user history
Strategic timing and future developments
Apple's cautious approach extends to its public communication strategy. The company isn't expected to discuss the Knowledge project during the upcoming Worldwide Developers Conference on June 9, 2025. Instead, the WWDC will likely focus on design innovations rather than AI advancements.
This measured approach stems from lessons learned from previous premature announcements, particularly regarding Siri and Apple Intelligence. By avoiding public commitments until the technology is fully ready, Apple aims to manage expectations more effectively.
The development timeline remains flexible, with no confirmed public release date for the chatbot. Apple continues refining the technology internally, addressing the hallucination issues while building a foundation for deeper AI integration across its product ecosystem.
As competition in the AI assistant space intensifies with advances from OpenAI, Google, and Anthropic, Apple's approach balances innovation with its traditional emphasis on reliability. When the Knowledge project eventually reaches consumers, it will likely represent a significant leap for Siri's capabilities, potentially reshaping how users interact with Apple devices.
