The Financial Times has revealed that Apple has hired a number of Google’s former AI experts to shake up the segment.
“That’s why Apple has been so careful”
And the project they have been working on since 2018, if not earlier, reveals the FT which has snooped on LinkedIn profiles and revealed that 36 AI specialists have gone to John Giannandrea (who himself worked at Google before being hired by Apple in 2018) AI team in Cupertino. And this is strange, because it is claimed that Apple already spoke in late 2013 about the possibility of making Siri part of a “neural architecture for speech recognition.” This according to Chuck Wooters with AI and LLM as expertise – he worked at the company for two years.
“Even back then, before big language models took off, they were big proponents of neural networks,” claims Wooters, who makes us wonder why Siri hasn’t gotten significantly better over the years, and why exactly Apple’s assistant didn’t become AI-smart first .
Ruslan Salakhutdinov from Carnegie Mellon University believes that the problem is that AIs can give incorrect or “problematic” answers, and that this is the reason for the extra caution. “They can’t launch something they can’t control,” claims Salakhutdinov, who knows a lot about the topic: his company, Perceptual Machines, was bought by Apple in 2016 with AI image generation as an expert field.
With iOS 18, Apple will probably present ways in which they control the AI, and how users can report errors.
