Apple Intelligence can still save AI—by solving the smallest problems

If you hoped flipping the calendar over to a new year meant that we’d stop talking about AI, well, you better keep flipping for a while. Eventually we won’t be talking about it because AI will be in charge of everything.

As my colleague Jason Snell recently wrote, prepare for 2025 to be the year of Apple Intelligence all over again. As we wait for the last tranche of features announced by Apple last year, the company will no doubt be ramping up to talk about the features it plans to add this year. All of this remains a contentious issue in tech. It seems as though nary a day goes by without some company or other announcing a new AI-powered lawnmower or weird keyboard. Right now the prevailing opinion in the tech industry seems to be that AI is the solution to every problem you encounter–and more than a few that you don’t.

While I think that’s far from true, there are places that AI–in particular, Apple Intelligence–could prove itself useful in our lives. For every Image Playground that has us wondering “who asked for this?” there’s a corollary of “why hasn’t AI solved this yet?”

Spamaheckofalot

After all this time, how is it that spam is still a problem in 2025? We’ve all just accepted that our email inboxes and text messages will be chock full of the stuff, but Apple Intelligence could actually help us. Apple’s made the barest of tips of the hat towards this with the mail categorization feature in iOS 18.2 (which, it should be noted, is still not available on either macOS or iPadOS), but that system is not only not specifically designed for spam, its efficiency at even categorizing mail is…hit or miss.

Here’s one big problem: iOS/iPadOS has no email spam filtering. Yes, there’s a Junk mailbox, but that’s reserved for messages you manually move to it; messages moved there by spam filtering on a Mac, if you have one; and messages categorized as spam by your email server. That’s quite the patchwork quilt with a lot of holes in it. At times, Apple’s spam filtering feels like three outfielders converging only to let a baseball drop between them. I still end up with spam in my inbox and good messages in my Junk mailbox–and let’s not even get started on iCloud’s silent email filtering which sometimes just doesn’t deliver messages at all.

And spam has encroached elsewhere too. Text messages and iMessage have seen a decided uptick in junk too, but calling Apple’s tools there “lackluster” is an insult to things that lack luster. There is no message filtering (beyond unread and unknown contacts), and Reporting Junk has questionable utility (it doesn’t even block the sender, for example).

Recognizing patterns is one thing that AI is actually pretty good at. Apple Intelligence should be taking on the spam problem across the board, because the end result should be that users spend less of their time managing their inboxes.

Foundry

Search and ye shall find (maybe?)

From Sherlock to Spotlight, Apple has a long history of trying to improve search on its platforms. Unfortunately, it’s still not good enough. A Spotlight search for “files edited in the last 30 days” does nothing. Even typing to Siri in macOS 15.2 gamely suggests I should try ChatGPT instead, which is an abrogation of responsibility to the extreme. Yes, you can set up an advanced search in the Finder and select all the right criteria, but natural language parsing seems like table stakes in this day and age.

Even basic search in some places seems woefully broken. Take Settings on iOS, for example: if you want to search for a particular section, say, Messages, you’ll find yourself wading through dozens of irrelevant results to find the one you’re looking for, if you can find it at all.

I don’t necessarily want Spotlight or Siri to provide me with all world knowledge at my fingertips, but in the domain of my files and information, it seems like this system should be better informed than any other source. Siri Suggestions on iOS has demonstrated that Apple’s systems can be good at anticipating their users’ needs, so a little bit of that old Apple Intelligence wouldn’t go amiss when I’m just trying to find that document I was working on yesterday.

Call me Dan the Automator

One of the places that I’ve used AI to great effect in recent weeks has been in working on some hobbyist programming projects. Programming is largely a bound problem, one that there’s a lot of training material for, and tools like ChatGPT and Claude understand it much better than I can with the time available to me to learn. And that’s not even talking about tools like Copilot or Apple’s Swift Assist that integrate directly into development environments.

Foundry

Me, as I said, I’m a hobbyist. But I do love automating a task to save myself time. Still, all too often, I have an idea, open up Shortcuts, and fumble with where to even begin. This seems like a place where Apple Intelligence could notice that I do a certain task a lot and suggest some steps I could take to automate it and save me time in the future. Is that complex? No doubt. But Apple’s systems should understand their capabilities and how to help users get the most out of them.

That’s really the case across the board with Apple Intelligence. Where these technologies work best for Apple is when they make its devices more capable (and more time-saving) for the people who use them. The late Steve Jobs was fond of describing computers as a bicycle for your mind, and I for one would like to spend less time fiddling with Allen wrenches and more time enjoying the ride.