Apple's Pragmatic AI: A Smarter Approach to Everyday Integration
At its recent Worldwide Developers Conference, Apple announced a set of new AI models integrated into their existing products. The company’s announcement stood out from the approaches taken by other tech giants.
Currently, OpenAI focuses on building ever-larger models aimed at creating artificial general intelligence. Microsoft is integrating various Copilot models into its Azure platform for enterprise use, fitting OpenAI's models into its business framework. Similarly, Google is developing increasingly large models to enhance its search engine.
Despite its relationship with OpenAI, Apple is diverging from the trend of building massive models. Instead, Apple is focusing on smaller, more specialized models that integrate seamlessly with existing phone functionalities. Rather than relying on massive general-purpose systems with uncertain accuracy, Apple’s approach involves using data already on the phone, integrated with language models to enhance system fluency.
Smaller models offer several advantages: they are easier to build, require less time, money, and energy, and can run on smaller devices. This approach provides greater flexibility in deployment and usage, avoiding the challenges of a single, all-encompassing model.
We have witnessed the Gartner Hype Cycle's peak of inflated expectations, and now seem to be entering the trough of disillusionment. Ignoring the depressing name, the trough of disillusionment is really the moment in which we start to take a technology seriously. Apple has learned from others' experiences: bigger models, while impressive, may not be as practical. For instance, OpenAI has developed a model that can pass medical exams, but such capability is not generally useful in everyday life. In contrast, Apple aims to integrate AI with everyday tools, like the calendar, offering practical benefits.
We are at the onset of a new era of integration, promising significant advancements for AI and language models. This includes upgrading Siri to provide more fluent interactions. By leveraging Siri's strengths and integrating it with other data, Apple’s approach allows for more accurate and relevant responses.
The positive reaction in Apple’s stock reflects the market's recognition of this strategy. While Google excels in search and Microsoft has an extensive product ecosystem, Apple’s strength lies in user experience and design. This approach is a testament to that expertise.
Kristian Hammond
Bill and Cathy Osborn Professor of Computer Science
Director of the Center for Advancing Safety of Machine Intelligence (CASMI)
Director of the Master of Science in Artificial Intelligence (MSAI) Program