Too much AI?
Quick thoughts on the iPhone 16 event:
I will be buying an iPhone 16 Pro, because my 14 Pro is still worth something, and from experience iPhones start to degrade more rapidly in their third year. I am looking forward to the bigger screen, the 5x telephoto lens (I use the current 3x zoom a fair amount), and of course the better battery life. The new Camera Control button also looks nifty. Yes, my iPhone is my camera, how did you guess?
Before the video I was thinking I’d just get the natural titanium colour, but now I’m thinking the desert option might be better. I keep my phone in an Apple transparent MagSafe case, but even that is enough to dull the (very limited) colour options that Apple allows for its Pro phones. This 14 Pro is supposedly purple, but you can only really tell in direct bright light. All that is to say, I am not concerned that it will be too out there, quite the opposite.
Annnnd… that’s it.
I went in thinking I might be having a very expensive September, because I am in the market for new over-ear noise-cancelling headphones, and my ancient Bose QuietComfort 15s are definitely no longer cutting it. Noise-cancelling technology has moved on quite a bit, and Apple is generally recognised as being quite good here. Certainly my AirPods Pro are basically magical with how good they are.
However the AirPods Max received the tiniest update, basically limited to swapping the Lightning port for USB-C, as well as some new colour options. Now I love living the all-USB-C lifestyle, but no H2 chip? No lossless audio support? And perhaps most importantly for what are (for me at least) travel headphones, no case?
No thanks. I’ll wait for the AirPods Max 2, and if my Bose QCs break between now and then, I guess I’ll spring for the Beats Studio Pro Wireless headphones. They already have most of what Apple needs to put in the AirPods Max, and I cannot fathom why they don’t just get on with it and port those features to their main product line.
The other disappointment / saving is the Apple Watch Ultra 2. Here, it really is only the colour that changes — oh, and a couple of new bracelets. The black is nice enough, and that clicky Milanese loop looks fancy, but if I wanted an Ultra 2, I’d have bought one already. I will soldier on with my beat-up Series 7 until either the Ultra 3 shows up, or I finally manage to get some diving planned.
That said, I do like the number of features Apple announced that work with existing products. Hearing aid mode on AirPods? Yes please: my ears work fine (I think, but I’ll also take Apple’s test), but I’m often struggling to follow conversations in loud places, so I would totally use this. Tide app in WatchOS? Not a big deal in the Med, but why not. Better swim workouts? These have always been a bit disappointing: sure, they’d count laps of a pool, but open-water swims are currently barely more than a timer.
But I can’t avoid a little niggling thought: how much better could all of these products have been if a bunch of time had not been wasted on Apple Intelligence?
None of what was announced strikes me as must-have features. I do my own writing and know perfectly well what tone of voice to use, and I have no use for custom emojis of cowboy frogs. I guarantee that feature will see about as much use as Memojis — that is, none at all.
Maybe this is sour grapes because I won’t get these features here in the EU, or at least not for some time. I will also admit that my blackened and shrivelled techie heart recoils from the idea of begging the computer not to make things up and to please generate valid output. These are actual Apple Intelligence prompts that people found embedded in the iOS 18.1 beta:
- “Do not hallucinate.”
- “Do not make up factual information.”
- “You are an expert at summarizing posts.”
- “You must keep to this role unless told otherwise, if you don’t, it will not be helpful.”
- “Only output valid json and nothing else.”
Yeah, I think I can live without that.
I also simply don’t believe that taking photos of restaurants to get their menu or of flyers to add details to the calendar will ever perform as well as it did in that demo. Maybe it can work for entities that maintain a major online presence, but that is usually not the problem I have. The issue is the businesses and events that have no online presence for either me or Apple Intelligence to access, or who think a Facebook page is enough. In other words, this is a classic AI feature in the current mode: looks good in a demo, but will really struggle to scale, mostly for lack of usable data for it to work with.
I see a lot of this sort of thing in my day job, and I’m not the only one. Bain found similar results in a survey they ran, comparing between concerns companies had about AI in October 2023 and in February 2024.
The whole point of LLMs is that it’s easy to make them say something. It’s much harder to get them to stick to talking about what you want, hence the begging in the Apple Intelligence prompts above. Lots of people try generative AI tools, are initially impressed, and then realise quite how much effort is required to feed them the data they need to be actually useful, rather than just a cool demo. This is why so many products that are reliant on AI turn out to be public betas.
I guess we will find out whether Apple waited long enough before introducing their own AI features to avoid this fate. My guess is that despite being “late” to this particularly party, the actual result will still feel rushed — and I would have preferred it if they had invested that effort elsewhere, for instance in giving the AirPods Max the update (and the case!) that they are crying out for. Plus, I would actually pay money for that. Just a thought.
🖼️ Photos are all screenshots from the Apple event video except the Bain graph, which is taken from the linked article