Apple is reportedly building a conversational AI that will take advantage of large language models to enable Siri in iOS 18 to automate multi-step tasks.
According to a new report by The Information, the primary goal is to enable the user to perform complex tasks that require multiple apps and services just with their voice instead of manually building custom automation in the Shortcuts app.
For example, this would let Siri create a GIF using a series of photos and send it to a friend via iMessage. It sounds like Siri will create custom automation based on the query using the technology powering Apple’s Shortcuts app.
Rumor: iOS 18’s Siri could automate complex tasks
The article suggests Apple plans to launch those Siri advances in iOS 18 next year.
Of course, OpenAI’s ChatGTP has popularized and mainstreamed large language models. But the truth is, these algorithms have been around for many years.
Siri’s been able to perform basic tasks on behalf of its users since its 2011 debut. Still, the assistant cannot execute queries involving multiple commands.
A large language model would enable Siri to parse such complex queries into a series of simpler actions that could be chained together. In other words, the underlying technology for such a feature already exists. All that’s needed is a more intelligent, conversational voice assistant that could parse multi-step commands.
The article goes on to note that John Giannandrea, Apple’s head of AI poached from Google, formed a team in 2019 to create conversational AI using large language models, which apparently costs the company “millions of dollars a day.”
Siri in iOS 17, iPadOS 17 and macOS Sonoma brings a few new features.
There’s a shorter “Siri” invocation instead of the usual “Hey Siri.” The voice assistant can also respond to multiple requests back to back without needing to reactivate it. Siri also powers rich visual results available in the iPhone’s StandBy mode.