Does Apple really have an AI problem? | Current affairs

Bloomberg Originals recently ran a story about AI and Apple. It brings up loads of valid points, that is if you think that AI - LLMs in particular - are 'the future'. It's curious that the publication doesn’t tap into the myriad of troubles around LLMs themselves.

Seeing the privacy backlash companies like OpenAI, LinkedIn and Meta face because they’re mining customer data for free in sneaky ways, I don’t think Apple would be wise to even entertain the idea to use their customer data for LLM training in the way the others do.

Take that privacy stance away and Apple stands to lose all credibility with their user base. Something they can’t afford. It’s literally part of the moat. The other part being their hardware and chips.

After all, modern smartphones are pretty similar because it’s a mature product category and market. Much to the chagrin of tech reviewers who don’t have much spectacular new year-over-year features to show.

Another reason is the backlash around the cut Cupertino takes from developers. While it was revolutionarily low back in ‘00s, it soon became the new norm. And when you’re completely dependent on the App Store, that cut can be pretty irritating and led to some legal battles and online upheaval, most notably around Epic Games. Whatever side you stand in that battle, it damaged Apple’s brand.

Add to that the rather panicky focus switch to AI for the 2024 releases of their operating systems. That always seemed more like doing a song and dance for shareholders, who demanded to see them doing something with AI. All those features they were promising weren’t realistic. Everyone with some knowledge about LLMs would know. So the retreat was always going to be painful and got some scathing critique.

I doubt the AI hype bubble will last long. Meta abruptly froze their recent hiring binge. OpenAI CEO Sam Altman talked about the bubble bursting. Even in a normal hype cycle, with a stable geopolitical and governing environment in the US, it would burst pretty soon.

I don’t see LLMs go away. More like Microsoft, Google, Apple and Meta ending up buying up OpenAI et al once the investors get cold feet and stop funding companies who realistically can’t turn a profit anytime soon.

The energy costs are immense, getting quality training data is very hard and LLMs can’t be used for reliable automation like regular ML.

My bet? Apple is (wisely) sitting this one out while at the same time make enough noise about doing something with AI to please the shareholders. Once the bubble has burst, investors will want to see something new again. And by that time, they'll need to be seen doing something with that.