According to the New York Times, Apple is testing generative artificial intelligence (generative AI) technology, which may one day be used in the Siri virtual assistant, although there are some fundamental problems in the way Siri is designed. question.
Apple reportedly introduced employees to Apple’s large-scale language model and other artificial intelligence tools at its annual artificial intelligence summit held last month. Apple engineers, including members of the Siri team, are testing the concept of language generation “weekly” in response to the rise of chatbots like ChatGPT.
These next-generation AI technologies underscore how Siri, Alexa and other voice assistants have squandered their lead in the race to artificial intelligence.
Siri, in particular, faces multiple hurdles that prevent it from making meaningful improvements.
John Burkey, a former Apple engineer who worked at Apple and was involved in Siri’s development and was responsible for improving Siri’s performance in 2014, said Siri was built on “clumsy code” and it took weeks to add even basic functions.
Siri’s “cumbersome” design made it difficult for engineers to add new features. Siri’s database, for example, contains so many phrases in nearly two dozen languages that “it’s like a big snowball,” Burkey said.
That means simple updates like adding new phrases to the dataset also require rebuilding the entire Siri database, which can take six weeks, while adding more complex features like a new search tool can take up to a year.
Therefore, in Burkey’s view, Siri has no way to become a “creative assistant” like ChatGPT.
DigiTimes also reported last week that under the catalysis of ChatGPT, people’s interest in generative artificial intelligence is increasing, prompting major companies such as Apple, Meta and Amazon to reevaluate the way the technology is developed.