Apple and Google: New details on the AI deal

Apple is further expanding its position in the field of artificial intelligence. New information reveals that the AI deal between Apple and Google is significantly more comprehensive than previously thought. The focus is on access to Google's Gemini models, their customization, and concrete implications for upcoming features like Siri in iOS 27. The details provide a clearer picture of how Apple is currently shaping its AI strategy.

The collaboration between Apple and Google seems contradictory at first glance, as both companies compete in many areas. At the same time, it demonstrates that modern AI development is increasingly based on cooperation. Apple strategically uses external technologies to accelerate the development of its own systems without abandoning internal research.

Scope of the AI deal between Apple and Google

A recent report describes the partnership between Apple and Google as "deeper than previously known." According to the report, Apple is granted significantly more freedom in its use of Google's technology than originally anticipated.

Specifically, Apple has unrestricted access to the Gemini model within its own data centers. This access is not limited to mere use; Apple can actively further process and adapt the technology.

Use of Gemini in Apple's infrastructure

Access to Gemini allows Apple to develop its own AI models based on this technology. The primary goal is to create specialized and smaller models that efficiently perform specific tasks.

These models can be optimized to run directly on Apple devices. This reduces reliance on cloud processing and ensures faster results in everyday applications.

Model distillation as a central method

A crucial component of this strategy is the so-called model distillation. This is a process in which a large language model transfers its knowledge to smaller models.

These smaller models are specifically designed for:

to require less computing power
to work faster
to be run locally on devices

The goal is to make the performance of large models like Gemini available as efficiently as possible in a compact form.

How smaller Gemini models benefit

By having full access to Gemini, Apple can not only adopt the results, but also understand and replicate the model's internal calculation processes.

This means that the smaller "student" models learn how Gemini arrives at its answers, and not just what the answers are. This approach leads to a significantly better quality of the derived models.

The result is compact AI models that come close to the performance of large systems, but require significantly fewer resources.

Technical challenges in implementation

The model distillation process is considered technically demanding. One problem is that Apple's goals for Siri don't always align with Gemini's strengths.

This makes adapting the models more difficult and necessitates additional development work. It's not just about performance, but also about consistency and practical applicability in everyday life.

Apple's internal AI development continues

Despite its close collaboration with Google, Apple continues to rely on its own AI models. The so-called Foundation Models team is still working on internal solutions.

However, it remains unclear what specific goals this team is currently pursuing. The report suggests that Apple is further developing its own AI strategy in parallel, without relying entirely on external technologies.

New features for Siri in iOS 27

A key application area of the new AI technologies is Siri. Apple is planning extensive changes, which are expected to be presented at WWDC in June.

The expected improvements include:

the ability to remember past conversations
a better understanding of context
proactive suggestions based on situations

One example of this is the ability to automatically suggest leaving earlier to avoid traffic problems on the way to an appointment, such as when being picked up at the airport.

Apple's AI strategy is changing

The AI deal between Apple and Google demonstrates how significantly Apple's strategy in the field of artificial intelligence has evolved. Access to Gemini, combined with model distillation and its own research, forms the basis for more powerful and efficient systems.

Siri, in particular, is showing clear signs of further development. At the same time, Apple remains flexible by both utilizing external technologies and further developing its own models. This combination is likely to be crucial in determining how competitive Apple remains in the AI field in the future. (Image: Mojahid_Mottakin / DepositPhotos.com)