App stores defined mobile. And, increasingly, conversation is defining how people start tasks with AI. — Bloomberg
For years, Apple insisted chatbots weren’t the way people should interact with AI.
Its position on what it thought customers wanted from artificial intelligence wasn’t exactly subtle. In an interview just last year, Craig Federighi, Apple’s senior vice president of software engineering, said the company never wanted to send users “off into some chat experience in order to get things done.” Apple’s philosophy, he argued, was that AI should be quietly woven throughout all of a device’s features.
I suppose that makes sense. Apple is uniquely positioned to offer a personalised version of AI that surfaces information based on your context and the apps you regularly use. The thing is, people actually like using chatbots. There’s a reason ChatGPT is the fastest-growing app ever.
Apple seems to have figured that out and changed its position.
The next version of Siri
According to Bloomberg’s Mark Gurman, Apple plans to rebuild Siri into a full-fledged, system-wide AI chatbot later this year. Internally code-named Campos, the new Siri will replace the existing interface and be deeply embedded across iPhone, iPad, and Mac operating systems. Users will still summon it the same way – by saying “Siri” or holding a side button – but the experience after that will look far more like ChatGPT than the assistant Apple has spent the last decade downplaying.
I think it’s fair to say that Apple has finally conceded what nearly every AI company has already learned: chatbots are the interface.
For a while, Apple could afford to resist that conclusion. Siri existed, but it was never positioned as the primary way people interacted with their devices. Mostly, that’s because Siri wasn’t very good. Yes, it would set reminders and start timers, but it has been pretty bad at almost everything else for a while now.
Conversational AI, meanwhile, was framed as inelegant and inefficient. Why type out a request when you can just use an app? Why build a system that allowed people to have a conversation with their device when the device could just know what they wanted because it already had access to all of their personal context?
Apple is playing catch-up
The answer is pretty obvious – the latter is much harder than anyone thought. And, again, chatbots turned out to be a lot more popular than Apple was willing to acknowledge. For all the ways Apple is behind in AI, the fact that it stubbornly held onto a belief that it was going to have a better idea might have been its biggest mistake.
Products like OpenAI’s ChatGPT didn’t just introduce increasingly better models; they retrained people. Users learned that software could handle messy prompts, follow-up questions, and mid-thought revisions. They didn’t need to know where a feature lived or which app owned a task. They could just ask questions.
Apple resisted that shift longer than almost any of its peers. As Gurman notes, even as Samsung, Google, and others embedded conversational AI deeply into their operating systems, Apple continued to argue that users preferred AI “woven directly into features” rather than standalone chat experiences. Campos represents a quiet reversal of that belief.
The AI interface
According to Bloomberg’s reporting, the new Siri will be capable of searching the web, generating text and images, summarising information, analysing uploaded files, and –crucially – understanding what’s on the user’s screen in order to take action across apps. It will be integrated into Apple’s core software, including Mail, Photos, Music, and even Xcode, allowing users to complete complex tasks with voice or text alone.
And that’s the point.
If we’ve learned anything about technology, it’s that interfaces define the experience. Browsers defined the web. App stores defined mobile. And, increasingly, conversation is defining how people start tasks with AI.
Gurman says that Apple isn’t launching a standalone chatbot app. Rather, it’s rebuilding Siri as a system-wide interface. Apple isn’t trying to compete with ChatGPT as an app. Instead, it’s trying to keep users inside Apple’s ecosystem, on Apple’s terms.
There’s an irony here. Apple’s long-standing instinct – that AI should be integrated, not siloed – wasn’t wrong. What it missed was that the conversation itself had become the integration point. Chat is how hundreds of millions of people are experiencing AI.
Figure out you’re wrong quickly
We should know relatively soon whether Apple is up for the task of building a capable chatbot. Specifically, the challenge is whether the company can make one that feels unmistakably Apple – personal, reliable, and integrated into how you already use your device.
Of course, the lesson here is that sometimes you’re going to get it wrong. Your goal should be to figure out that you’re wrong as quickly as you can so you can move on to whatever you should be doing instead. In Apple’s case, we should give the company credit for realising that its original way of thinking about AI was – at the very least – incomplete. There’s plenty to unpack about why Apple got AI wrong, but it’s a very good sign for all of us that the company is willing to change course. – Inc./Tribune News Service
