The thing with Apple is, up until the Apple Intelligence debacle, they were renowned for being a company that would hold off on new technologies and then release the polished version everyone actually wanted.
They could have, and I feel should have, held off on the launch of Apple Intelligence for a few more years. ChatGPT was only released to the market in November or December of 2022. At present, the tech world is still grappling with the limitations of LLMs and how to best use them. Even Google, supposedly a titan of services, is wildly flailing about and largely failing.
I say this as someone who uses the ChatGPT API daily to massively increase work efficiency: LLMs are hugely flawed and are potentially dangerous for the kinds of things people think they can use them for.
The reason I can use LLMs well is because I know their pitfalls and don't trust them at all. I am also an expert in my field and know how to apply them in the right way. I also have a multi-step quality assurance process before client delivery.
Despite the hype, LLMs are not ready for consumer-facing tasks. Just the other day I got a summarised e-mail notification that a huge sum had been debited from my account. "Yeah, right," I thought, and when I opened the e-mail, surprise surprise, it was more like a few hundred dollars; Apple Intelligence had done the math wrong.
I've seen hallucinations that were just as bad (or worse) even on OpenAI's most advanced models.
I wonder whether the Apple executives have really gotten their hands dirty with LLMs and have seen just how broken they are. I don't understand why they panicked and pushed Apple Intelligence to market.