| "When we think about what it means for intelligence to be really useful, it has to be centered on you," said Apple's senior vice president of software engineering, Craig Federighi. "We think AI's role is not to replace our users but to empower them. … It needs to be integrated into the experience you're using all the time. It needs to be intuitive, but it also needs to be informed by your personal context. Really, knowledge of you."
Apple is creating a personal kind of AI experience in a number of ways, including a new and improved version of Siri that is contextually aware of your data. One example of how this will play out that Apple shared is about having to pick up your mother from an airport.
Ask Siri when your mom's flight is landing, and the assistant will find the appropriate messages your mother sent you about her flight via email or text, then check real-time flight data to give you an exact time when her flight lands.
Siri will also be able to understand the content on your screen. Say your friend sends you an address followed by a series of unrelated texts. You can tell Siri, "Save this address to their Contact Card," and Siri will recognize that you want to save that particular address to your friend's Contact Card without having to say your friend's name over again.
Siri isn't the only place you'll find Apple Intelligence. Apple says it's using the technology to help prioritize notifications, putting priority notifications at the top of your notifications stack and even summarizing long threads to help you catch up without having to scroll through endless messages.
Like Google and Samsung, Apple has also added generative AI-powered photo editing features and the ability to summarize and rewrite text in Mail and other apps.
But generative AI raises privacy concerns since data is regularly shared back and forth with the web. And it could be especially problematic for Apple to intertwine so much of your personal data with generative AI apps. To that end, Apple says it is using a number of on-device AI models that don't share your data with cloud-based services.
For requests that are too complex for on-device models, Apple says it will connect to its own specialized cloud services it calls Private Cloud Compute, which runs on Apple's own servers using its own chips, which the company says ensures requests remain secure. What's more, the company says that when users make requests, only a small amount of necessary data goes up to the cloud.
Apple added that its servers have no storage option, so it won't save any of the data from your requests. The software powering Private Cloud Compute is also publicized so that it can be audited by security researchers.
Basically, the company wants users to know that while some AI prompts will use cloud-based AI models, your user data won't end up being used to train future Apple models or live on Apple's servers in perpetuity.
Apple isn't only using its own AI models. It has also partnered with OpenAI to bring ChatGPT to the company's devices. But Apple says you'll be prompted as to whether you want to use the service before anything is sent to the OpenAI's servers. Apple says it's also working to integrate other AI models including Google's Gemini.
Apple Intelligence will be available on the iPhone 15 Pro and Macs and iPads running its M1 chip and newer later this fall. And if the company is successful in its effort to make generative AI truly useful for the average user, it could set Apple up as one of the industry leaders in the still-nascent space. | 
No comments:
Post a Comment