### Behind the Curtains of Apple’s Recent AI Departure
Recent findings suggest notable exits from Apple’s AI division, sparking worries about the company’s trajectory in the realm of artificial intelligence. A comprehensive analysis by *The Information* uncovers internal disputes, especially regarding the possible open-sourcing of Apple’s AI models and the repercussions of such a decision.
Earlier this year, the team dedicated to Apple’s AI models contemplated the open-source release of several of them. This effort aimed to highlight Apple’s innovations in AI while utilizing external research to refine the models. Nonetheless, worries emerged concerning the performance decline of these models when adjusted for iPhones as opposed to their more robust counterparts in data centers.
Craig Federighi, Apple’s software head, opposed the open-source plan, contending that existing models from other firms already fostered research. He expressed concerns that unveiling Apple’s models would expose their weaknesses against rivals like Alibaba and Google, leading to a public perception that Apple was sacrificing its software quality to adhere to its on-device strategy.
This internal conflict has allegedly played a role in the recent departure of AI experts from Apple. Numerous researchers assert that the company’s dedication to an on-device AI approach is considerably restricting the performance of its models. Federighi himself admitted this issue in conversations with Ruoming Pang, the previous leader of the foundation models team.
The introduction of Apple Intelligence last year highlighted user privacy, adopting an “on-device first” mindset. However, this tactic may be constraining the potential of Apple’s AI features more than the company has openly recognized.
Additional findings from the report include:
– Apple’s AI researchers were taken by surprise by delays in Siri’s feature launch, having earlier received favorable evaluations of their efforts.
– There are conversations about Apple possibly engaging third-party AI models rather than depending solely on in-house creations.
– Discussions have occurred with OpenAI, Anthropic, and Google regarding the adoption of their large language models (LLMs) to improve Siri.
– The remaining members of the foundation models team have been notified that Apple is reevaluating compensation to promote retention.
These insights offer a window into the obstacles Apple encounters in the competitive AI environment and the strategic choices that may influence its future course. For further information, the complete report is accessible [here](https://www.theinformation.com/articles/apple-losing-ground-ai-talent-war-just-money).
