Unpacking Apple Intelligence’s several layers, actual Zoom fatigue, and dubious VPNs
Apple Intelligence’s: There is still more work to be done before it is clear how precisely Apple intends to implement Apple Intelligence.
Apple Intelligence’s : Though iOS 18, iPadOS 18, and macOS Sequoia are expected to roll out in late September (typically after the iPhone keynote, in time for the availability of the new iPhones), there is still some work to be done before it is clear exactly how Apple plans to push out Apple Intelligence. First off, there are hints that the new operating systems would not be released all at once; rather, availability might be made available in phases throughout a longer testing phase. That seems improbable to me because that isn’t how Apple asks customers to join a waiting list for new software features. However, there might be chapters on ChatGPT integration, Siri enhancements, and Apple intelligence. Apple has not yet commented on availability; everything you read online is merely conjecture disguised as gravity in order to sound, well, serious. To me, acceptability isn’t Apple’s main problem. It will be to make sure the generative AI models, which are notoriously unpredictable, don’t go crazy.
One thing became evident from my series of discussions with Apple executives outside of the WWDC keynote: the tech giant’s internal efforts are behind the Apple on-device and Apple server models. not GPT-4o from OpenAI, which is actually only one parallel, optional component. That, as far as I can tell, has no bearing on how Apple’s internal AI models function, develop, or react. You’ll be asked to confirm if GPT can be used if there are any instances in which your inquiries might need more in-depth solutions that involve specialized knowledge. It will be up to you to decide whether or not to accept. Having said that, there are a few things Apple should be concerned about because it is working with an unpredictable and erratic entity. A checklist is provided here.
In particular, generative AI is only truly effective when it has a large amount of data at its disposal. Even then, there’s no assurance that it won’t contain errors—something we’ve repeatedly witnessed.
Walking a tightrope with AI means that Apple must successfully complete activities that are difficult, maintain context that is layered, and personalize without jeopardizing data privacy, all while minimizing false positives, hallucinations, and the creation of potentially harmful information for people.
I was informed by Apple executives that the company will not use user data for model training; instead, training and context will be developed using more general activity trends. Attempting to compute as much as you can on the device is one method to achieve it.
Apple’s models won’t retail any user queries. So much so that they have OpenAI comply with the GPT-4o integration so that users may choose to call them for specific tasks. More specifically, OpenAI affirms that users’ IP addresses will be hidden and requests will not be saved by OpenAI while using ChatGPT through Siri or Writing Tools.
Time will tell just how this plays out. Managing the unpredictable generative AI presents Apple with its greatest challenge in recent memory. This goes beyond a chatbot. The system-wide integration of Apple Intelligence increases a user’s interactions with AI. No stress then!
Examining the AI chapter that Apple is writing right now…
Detailed examination of Apple Intelligence models, ChatGPT option, and privacy policy
Beyond Apple Intelligence, iPads and Macs are evolving along with iOS 18.
Important note: Apple executives have verified to me that your interactions with ChatGPT under Siri are the exclusive scope of the company’s agreement with OpenAI for GPT-4o as part of Apple Intelligence. However, the agreement, which guarantees that OpenAI will not save your data and would conceal IP addresses so that usage cannot be linked back to users, will not apply to you if you log into your OpenAI account using a paid membership for ChatGPT Plus.
SAFE?
Apple Intelligence’s
Not too much if you’re relying on one of the several free VPN apps for Android. The issue is concerning, as Top10VPN.com’s Head of Research, Simon Migliano, explained to me. Here’s a sample: up to 88% of these free VPNs experience data leaks of some kind (IP, DNS, WebRTC; of these, about 17% are affected by multiple leaks); approximately 71% of apps share your personal information with outside parties, such as Facebook, Yandex, and dubious data brokers, such as Kochava; up to 15% of apps contain Bytedance tracking code; and 53% of apps have at least one function in their source code that could potentially compromise user privacy and ask for permission to be executed. You should read his in-depth analysis here; you’ll learn who the worst violators are, and it should motivate you to take corrective action. This is really concerning because, on the other hand, VPNs are typically fairly expensive, and consumers typically choose for the more affordable options. Consider this: they can sell your data to the highest bidder, which explains why they are free.
CONFERENCES
Virtual meetings are one (among many) lasting consequences of the epidemic age. It is important to remember that this fleeting period of time in our life did impart one important lesson: working remotely may be just as productive as working in an office every day if the work is done diligently by real people. You’re saving time by cutting down on your commute, water cooler chats, and—for some—longer lunch breaks. The success of virtual work depends on the execution of virtual meetings. We appreciate the contributions from Zoom, Google Meet, Microsoft Teams, Cisco Webex, and everyone else who contributed an app. However, it’s likely that we’ve started doing too much of a good thing over time. And I’ve addressed that issue in my most recent piece for HT Wknd. I’ve included a number of research that tend to highlight how awful things can get if you spend too much time in virtual conversations. Virtual meeting weariness is starting to become a real phenomenon. As I’ve mentioned, Zoom’s AI companion for call summaries and Microsoft Copilot for Teams call transcriptions may be able to help. However, are enough individuals aware of these available tools? Consider it.
(Premium): Zooming in…and zoning out: Is AI able to alleviate video call fatigue?
GENERATION
There is only basically one way to categorize the Apple iPad Pro 2024 version: a return to form and Apple’s redoubled pursuit of tiny devices. Ideally, it ought to provide an insight into what lies ahead. As I stated in my review, the iPad Pro series is at a turning point, and this chapter is defining what a potent computing device may be. Given that it is unveiling the M4 generation—the first iPad to have that luxury with Apple Silicon—you would expect nothing less. The eye-catching feature of the new OLED screens—which are actually two OLED panels fused into one—may draw your attention, as it should, but there’s another element that might have more of an impact. In the next years, we may see even thinner MacBook Pros, MacBook Airs, and iPhones thanks to the slimmest iPad Pro ever. Apple succeeded extremely well with the iPad Pro 11-inch and iPad Pro 13-inch—no one saw this coming. Apple Silicon is at the center of this shift because the business has discovered the elusive balance in the performance per watt measures. In the upcoming versions of the MacBook Pro and MacBook Air, everything that has helped make the iPad Pro thinner might still be applicable. A larger screen for the “Pro” phones might free up space for slimming everything down (this year’s iPhone 16 may be too soon, I fear). However, since physics will present more obstacles, it may need more for the iPhone. It’s an intriguing prospect. Let’s wait for everything to work out as it should.