Apple Intelligence is Sherlock and Watson all over again

Apple Intelligence marks a turning point. Not just for Apple, which finally learned how to say “AI” and is diving head-first into making big generative AI features a central part of its products. It’s a turning point for the tech industry as a whole.

This year, AI begins its transition from a novel product to an expected feature, a move made by dozens of software technologies before it. And Apple Intelligence plays a key role.

The software technology lifecycle

It’s not true of every piece of software, but it happens a lot: something you used to have to seek out and buy eventually just becomes a baked-in part of the products you already have.

The most famous example of this for Apple users is Sherlock. That was the name for Apple’s system-wide Mac search feature introduced way back in macOS 8.5. A third-party company (Karelia Software) sold a neat $30 utility called Watson that added internet search and other capabilities to Sherlock. Users loved it.

Within a couple of years, Apple just added those features to Sherlock, making Watson obsolete. They never licensed or paid Karelia, didn’t buy the company, they just took what used to be paid software and turned it into a feature.

We now call this “getting Sherlocked” and it happens all the time. You used to have to buy software to burn CDs or watch DVD movies. Antivirus software wasn’t built into anything. Companies like f.lux got Sherlocked by Night Mode display settings. All the big tech companies do it.

From speech recognition to health trackers and so much more, software and services that you had to find, choose, and pay for eventually become included as part of the operating systems and devices you use. It happens on iPhone, iPad, and Mac, it happens on Windows, it happens on Android.

This year, Apple is being accused of “Sherlocking” 1Password and Dashlane by making Passwords its own app, Truecaller by adding native call recording and transcription, and Magnet with the new macOS Sequoia window tiling feature. And now with Apple Intelligence, it’s starting to do the same to gen-AI apps and services.

The transition will take years

Apple’s not alone here. Microsoft’s new Copilot+ PCs require powerful NPUs (Neural Processing Units) and offer a mix of on-device and cloud AI services that include image generation, live captions on any video, and a built-in chatbot interface that works together with outside AI services like ChatGPT. Google has been infusing Android with its AI over the last couple of years and this year’s Android 15 release adds even more generative AI with an upgraded multi-modal “Gemini Nano” on-device model, on-screen awareness, real-time scam call detection, and more.

Artificial Intelligence is a big field and still going through rapid growth, so its transformation from product to feature won’t happen in just one release. Just as people kept buying CD-burning software to get more powerful features for years after it was built into computer operating systems, stand-alone AI apps and services aren’t going to disappear anytime soon. People will want more than what is built in.

But 2024 is the start of the next phase; the phase where AI, like so many products and services before, reaches most people as a feature that is baked into their phones, laptops, and tablets. What is built-in will be “good enough” for most, and integration with outside services will be free and seamless.

Take Apple’s deal with OpenAI to integrate ChatGPT. Whether you’re using Siri, writing tools, or art tools, you’ll be able to call upon ChatGPT-4o’s expanded capabilities right from within the OS interface, without downloading an app, signing up, or signing in. Over the coming years, this will start to look like the search engine defaults in your web browser–most people won’t know or care that they can be changed or why they would want to. You just open your browser and search.

AI will eventually become boring

Apple makes improvements to its Spotlight system-wide search (the feature that grew out of Sherlock) every year. Nobody really cares, though. It’s barely noticed and not a selling point anymore.

The same can be said of all the other software and services absorbed by platforms over the years. They start as neat new features worthy of keynote stage time and advertisements. Eventually, people get bored with it. They become features we’ve had and used for years, and the improvements are welcome but not worth crowing about.

This is the obvious endgame for generative AI, though the timetable looks longer. We’ll hear about amazing new AI capabilities as part of our operating systems (from Apple and others) for a few more years at least. Features pioneered by other companies in their stand-alone apps will slowly make their way into “free” features built into our products.

And when there’s enough there, and it works well enough, to meet the needs of almost all customers, it’ll be just another boring feature that gets updated every year. A throwaway line from the CEO as he transitions to talking about the next big thing.

Generative AI as a feature is not really a thing yet. You can’t yet walk into Best Buy and pick up a Copilot+ PC, Apple Intelligence doesn’t roll out to millions of users until the fall, and Android 15 is going wide in the second half of this year. But by January 1, 2025, tens of millions of people will use products where generative AI is just a built-in feature.

That’s a huge shift in public perception. It won’t take long for hundreds of millions of users to just expect this to be the way things work, relegating self-selected and paid-for AI as a niche product for hardcore users with particular needs.

iOS, MacOS