Omdia view
Summary
At its recent Worldwide Developers Conference (WWDC), Apple pulled the curtain back on its new suite of AI-powered capabilities, creatively named “Apple Intelligence.” For months, industry commentators have been speculating about Apple’s plans for leveraging advanced artificial intelligence (AI) capabilities, including generative AI, within its ecosystem, and WWDC brought clarity around Apple’s approach. The new on-device and cloud-powered AI features will be incorporated into iOS 18, iPadOS 18, and macOS Sequoia. This report explores the potential of these capabilities, specifically relating to business operations and employee experience.
Apple’s AI strategy fuses on-device and cloud AI
Apple’s AI approach and strategy blends on-device AI with an OpenAI-powered cloud-based AI. This provides the level of general intelligence many will have become familiar with via ChatGPT, with the rich, personalized, and multimodal intelligence that can be gathered and processed at the device level. The combination of native, on-device, and cloud-powered approach that Apple has embraced will help make user experiences more personal and improve how people interact with technologies across the Apple ecosystem while also providing a degree of flexibility in how AI models can be secured. Apple’s on-device AI model is independent from its OpenAI partnership, meaning the processing of any data via Apple’s native on-device model is carried out via the device with no relaying of information or data over the internet without the user’s consent. When data does move from the device model to a cloud-based AI model, it is imperative that Apple is transparent about what data is being transferred, where it is being transferred, and how it will be used. Beyond the security benefits of local data processing, Apple’s on-device model will also benefit from the security provided by Secure Enclave, a dedicated security processor found across Apple devices that governs how other components use local data.
This bimodal approach to AI modeling is a major benefit of Apple’s overall approach, especially when considering enterprises’ data security and privacy needs. On-device processing of AI tasks is prioritized on the hardware itself, reducing the need for data to be transferred over the internet and, in turn, reducing the risk of data leakage. Other things to consider regarding on-device and cloud-based AI models include:
- On-device models are not trained on the same extensive datasets as their cloud-based counterparts.
- Cloud AI models can also scale up in terms of computational resources when required, enabling them to compute highly sophisticated tasks.
- The device hardware impacts the performance of an on-demand model. Silicon and devices optimized for AI processing deliver the best performance. Cloud-based models work across any device that has an internet connection.
- As on-device models do not need to relay information across the internet, users benefit from better speed and performance when interacting with AI.
- On-device AI processing is resource-intensive, which can affect device battery life. Advances in this area will continue to be important as AI becomes more richly embedded in the native mobile experience.
It is also feasible to speculate that, in much the same way that IT teams can currently manage on-device data for corporate-provided apps and other types of data via unified endpoint management (UEM) solutions, the same may eventually be true for data training on-device AI models across the corporate-owned estate. For businesses, this will provide a wealth of insight and information that they can use to improve employee experiences and corporate security. However, it would again be imperative that businesses are transparent about what on-device AI model data is being used and how. This would also not be a feasible approach across personally owned devices that employees use for work purposes, such as emailing and unified collaboration.
Personalization is power
Apple’s AI model richly infuses AI capabilities into Apple devices and the native OS experience across mobile and PC. Via a mobile device, AI models can leverage multiple modes of input, including text, speech, images, and video. With this level of input and data ingestion, the potential for Apple to deliver richly personalized experiences, notably via Siri, is huge. The value of Apple’s proposition will not only be in how it can reference these elements, but in how it will be able to potentially leverage that insight via the applications that people use regularly across their devices, enriching experiences as a result. One of the benefits of generative AI and natural language models is how they can improve the way that people interact with technology. Apple’s new AI capabilities will significantly advance Siri, making it a more reliable and personalized solution that people will use to interact with apps and OS capabilities in new and more powerful ways. Features including more accurate text-to-speech, granular interaction with in-app features, content summarizations, and intelligent photo editing are examples of how Siri will advance. As these features are device native, they will inevitably be used for both business and personal activities. Other ways AI-powered employee experiences can be advanced and made more personalized could include:
- Technology adoption: Getting employees to maximize how enterprise applications are used, and thus the value they deliver, can be challenging. Intelligent recommendations and natural language processing (NLP) will provide employees with new ways to interact with business apps and data. This will enable them to intuitively undertake a range of tasks and activities while also helping businesses realize more value from technology investments.
- Efficiency: By improving the way that employees can interact with business data, AI and NLP will also help people carry out tasks and make their workflow faster and more efficient. For example, Siri will be able to better manage the administrative tasks employees regularly need to undertake across different apps. The on-device AI capabilities will adapt to user behaviors over time, delivering more relevant work-based content and notifications.
- Cost: AI costs, including service subscription fees and computational/processing-related costs, are a major challenge for businesses to navigate. These costs are also a barrier to the democratization of AI capabilities. Apple’s on-device model will be richly embedded into the OS and will not incur a subscription fee.
- Customer service improvements: By having easier and quicker access to specific business data and insights, employees will also be able to deliver a better and more personalized service to customers.
- Technical support: Technical support can also be improved by intelligent on-device agents helping employees quickly troubleshoot device-related issues. Where issues need to be escalated, a more comprehensive list of actions that lead up to any issue can be automatically relayed to technical support teams via an on-device intelligent agent.
Democratizing AI one device at a time
The two biggest challenges businesses face in adopting generative AI are understanding value and use cases and democratizing AI capabilities to ensure it gets into the hands of as many employees as possible. Regarding the latter, technical and cost barriers are notable issues currently preventing greater democratization of the technology. Apple’s introduction of native, on-device AI features helps overcome the democratization challenge by providing people who own a relevant device with AI capabilities. With this level of scale, it is also important to consider the costs involved in processing all the data that will potentially be generated. For example, at the time of writing, OpenAI’s API pricing for GPT-4o is $5 for 1 million input tokens and $15 per 1 million output tokens. Unquestionably, Apple’s investment in advancing its AI capabilities is a significant one, especially from an infrastructure perspective.
Apple’s inclusion of AI at the device and OS layer also delivers a level of standardization that gives app developers a consistent template upon which they can design AI experiences and apps that leverage AI capabilities. This consistency means developers can create apps with clear knowledge and understanding of the AI models that can be referenced. Finally, this development standardization and broader adoption of AI will help usher in a wave of new consumer and business use cases—something that many organizations are desperate for. Enhancing its phone, tablet, and PC operating systems with AI will undoubtedly further consumerize AI capabilities. These features will enhance the native experience people have with their Apple devices while providing developers with new ways to innovate across app and hardware experiences. AI that is delivered as part of the native device and OS experience is powerful as it reduces barriers to adoption while providing developers with a consistent ecosystem upon which they can deliver AI-augmented experiences. Apple brings the development resources and ecosystems, as well as the hardware and software needed to deliver compelling business and consumer AI use cases.
Appendix
Further reading
“Generative AI smartphones will usher in a new era of business mobile innovation” (April 2024)
Apple, “Apple Intelligence Preview - Apple,” Apple Intelligence (retrieved June 11, 2024)
Author
Adam Holtby, Principal Analyst, Workplace Transformation
