Computers since their invention have been a tool that a user uses. It does nothing without the user controlling it. This is all changing and I thought it worth a post, as the gravity of this change and its irreversibility does not get enough attention. No longer is Golden Gate a simple upgrade it is a shift of control. As this shift is not all at once, but subtle, it is easy to miss. Too easy. For myself I needed to do a deep dive into the implications, and I would suggest you do too. Especially before you hit the āupdateā button when Golden Gate is released.
My point is that AI will be part of the architecture going forward that you will be unable to change, or turn off. Instead of you controlling your machine you become a manager and AI controls your machine. You can tell it what to do, but how it does it will be a black box. People will no longer manage their files, email, calendar and system. AI will manage it and be in control of it. This comes with some tremendous benefits. You can tell AI to find the quarterly figures for you project and create a nice chart. Or sort your calendar appointments for the next month etc. The downside is control and cost. You no longer control. Any glitch in AI can have catastrophic results. You also burn the bridge of reversing this. Once AI is architecturally inbuilt you cannot go back; probably why the AI off switch is no longer present in Golden Gate. Then you need to consider the ongoing cost increase. AI needs more power, disk space and RAM. As it improves so do these needs. Upgrade cycles for your hardware will not only cost more, but also be shorter.
What Iām saying is that simply updating and hoping for the best may not be the best decision. I would encourage everyone to properly think through the ramifications and the cost. This is a major architectural change to computing not just an update with a few new features and it is not reversible.
Things like Spotlight, predictive text, system wide tools and the like that used to wait for you to do something before they acted. This has been handed to AI who processes that data actively in the background trying to anticipate what you want. You cannot turn this processing off.
But how does that work long term? If AI controls spotlight so it can understand natural language or understand your writing style, or create context aware responses to email or text messages AI has to be part of the core system. This makes turning it off next to impossible. Itās not just about the features such as context aware responses, but the underlying data processing and algorithms that make that possible. This only works if it is deeply embedded.
Well, from one of those threads I see this, which makes sense:
Iād be more surprised if they did allow it to be disabled in early beta. After all, itās the most advertised flagship feature of the new OS, and one that is deeply integrated into core functionality. The whole purpose of the beta is to test new features and capture bugs. Canāt do that if itās disabled altogether.
IF - since you absolutely do have settings to turn off Apple Intelligence then it canāt be that deeply embedded in Core.
I get your bigger concern about the direction this is going, as it may be something where, in the future, you canāt opt out. But since it is a major new feature then you either embrace it for what it is or you opt out, or just donāt upgrade!
Personally I think the main issue is that the extra 27-30GB it requires might push upgraders to also upgrade their hardware at the new crazy memory prices. This probably wonāt include me, at least for awhile.
I agree, the AI scenario you describe seems like one very possible future. Especially when dealing with a āwe know whatās bestā corporation. But I suspect businesses will have a lot to say about on-device AI at some point.
I have to believe this has always been Appleās plan. But it isnāt mine.
My computing needs have always been pretty simple. My āworkflowsā were mainly on the servers, etc. that other people used. So while I am looking forward to Siri AI, most of my data will remain in the cloud. Iāve never been comfortable keeping a lot of information on my iPhone and iPad where it can be lost or stolen. Or inspected by someone when Iām traveling internationally.
At this point I donāt know how everything will work. My email, contacts, calendar, and primary task data is in my Google Workspace account, and my Apple devices just log in to use it. And I can limit the passwords I travel with using travel mode in my password manager. But it takes 40 days to purge messages from iCloud so Iāve already started deleting the thousands of messages that have accumulated in my account.
Iām probably going to need a separate iPhone, logged into a separate iCloud account, for some trips. I guess Iāll be dusting off my iPhone 11 at some point to see what works best for me.
Iām not saying itās mandatory (although you cannot toggle it off in the beta). I am saying that I believe Appleās trajectory is heading in that direction. IMO it will become increasingly difficult to toggle something that is increasingly becoming part of the architecture not just the interface.
Iāll still rely on cloud models for the majority of what I use AI for, but Iām happy to let Apple Intelligence + Siri surface stuff thatās on my devices.
The difference is the fundamental premise of the discussion - that computer users will āNo longer have control.ā I do not see any evidence for that.
I do understand what you mean about AI being part of the core system. But the question there becomes what āAIā means in practice.
I might be misunderstanding your point, but I donāt think AI ācontrolsā Spotlight any more than whatever Appleās current search back-end does. Thereās a query, thereās some sort of data store, thereās a persistent āworkerā process that builds and maintains that data store, and thereās an API or something similar that returns results based on parsing the query against the data store. Thatās how it is now, and itās how it would seem to be with AI.
If thatās what itās doing, AI isnāt controlling your computer any more than macOS has been doing for a decade or more. Different? Yes. But more? Not really.
Agentic AI is a different thing. Some of Appleās AI tech is agentic (the thing that helps you change your passwords on websites), but itās only launching the agent at your request. No request, no agent. And Iām not aware of anybody thatās seriously suggesting that we mandate the use of agentic AI.
From where I sit, AI is mostly letting your Mac do the sorts of things it already did, but better. As long as itās not uploading my tax returns to the cloud to harvest data to include in texts to scammers, Iām generally fine with it.
The evidence for that is that the beta version has no toggle to turn off AI.
I agree. The new Spotlight indexes a semantic, cross-app representation of personal data, not just separate per-app keywords. This representation isnāt human-readable, and its existence creates standing exposure risk ā to theft, compelled legal access, or future repurposing (including advertising) ā even if the user never queries it. It may run continuously in the background, and thereās current, real uncertainty about whether a master off-switch will survive into the public release.
That may perhaps be an issue of privacy. But that is very different from the initial stated premise that "You can tell it what to do, but how it does it will be a black box. People will no longer manage their files, email, calendar and system. AI will manage it and be in control of it. "
Yes, itās perhaps an overstatement at this point. But the direction of travel points to this being inevitable.
I should add that by ācontrolā I donāt mean AI dictates how you use it, but that it will capture and file the data rather than the user manually filing it.