Ex-Apple Engineer Explains How Siri Fell Behind in AI Race
Apple’s Siri has lagged behind competitors like Google Assistant for almost as long as it has been around, but it has more recently taken serious blows from emerging generative AI-based chatbots and assistants like ChatGPT.
In a recent interview with The New York Times, former Apple engineer John Burkey, who spent years working on the company’s virtual assistant, explained how it fell behind in the AI race.
Apple launched Siri in 2011, much to the same excitement and awe as OpenAI’s ChatGPT and similar technologies like Microsoft’s “new Bing” are garnering today. However, even after 12 years, voice assistants like Siri, Google Assistant, and Alexa haven’t evolved much.
According to Burkey, Apple’s lack of innovation with Siri was a result of the company constantly running into technological hurdles. The ex-Apple engineer, who was charged with improving Siri in 2014, said that the voice assistant’s code was inefficient, making it so even minor updates took weeks to push out.
From the beginning, Siri had a clunky design that made it time-consuming for developers to iterate and add new features. Where today’s AI chatbots can generate entire sentences and understand a wide range of inputs, voice assistants like Siri are better described as “command-and-control systems,” capable of only understanding a finite list of questions and prompts.
Siri is powered by a massive word database that comprises all of the requests and questions it can recognize, including the names of musical artists and real-world locations in multiple languages.
This made it “one big snowball,” said Burkey. If someone wanted to add a word to Siri’s database, he added, “it goes in one big pile.” The engineers working on Siri would have to rebuild the entire database, which could take up to six weeks, every time they wanted to add some new phrases. To top it all off, Burkey said that larger updates like adding new search tools could take almost a year.
While Google Assistant and Amazon’s Alexa relied on similar technologies under the hood, the former has usually bested Siri in side-by-side showdowns. All three, however, have been put on notice with the arrival of ChatGPT and Co. Earlier this month, Microsoft CEO Satya Nadella said in an interview that Siri, Google Assistant, Alexa, and even Microsoft’s own Cortana were all “dumb as a rock.”
Experts expect voice assistants and AI to eventually converge in the future, making modern generative AI’s natural communication, reasoning, and creative capabilities even more approachable with voice.
Apple isn’t about to sit this one out, either. The iPhone maker recently held an employee-only AI event where it showcased its own large language model and other AI tools, according to The New York Times‘ sources. Apple currently has several engineers, including those working on Siri, testing language-generating concepts every week.