A new report from The Information has shed light on the internal struggles behind Apple Intelligence’s revamped version of Siri. According to sources, Apple considered building both small and large language models for iPhone and cloud use but ultimately decided on a single model instead.
The decision was reportedly driven by conflicting personalities within the company and an overly relaxed culture. Poor leadership is said to be to blame for the execution problems, with some former employees citing a lack of ambition and appetite for taking risks when designing future versions of Siri.
Despite Apple’s AI chief John Giannandrea’s confidence in fixing Siri with better training data and web-scraping, senior leaders didn’t respond with urgency to the emergence of ChatGPT in 2022. In contrast, OpenAI’s technology outperformed Apple’s models.
Siri leader Robby Walker focused on “small wins” like reducing wait times for responses, but some efforts were shot down due to concerns over performance and emotional sensitivity. A project codenamed “Link” aimed at developing voice commands for the Vision Pro was also dropped due to technical issues.
The demo of Apple Intelligence’s most impressive features at WWDC 2024 was allegedly fictitious, with only one feature – a pulsing ribbon around the display edge – being activated on test devices. The decision to showcase an artificial demonstration marked a departure from Apple’s usual practice of showcasing working features and products.
However, some Apple employees are optimistic that Craig Federighi and Mike Rockwell can turn Siri around by allowing engineers to use open-source models and prioritizing AI development.
Source: https://www.macrumors.com/2025/04/10/chaos-behind-siri-revealed