I am particularly interested in how AI may be integrated into Apple’s default apps, as well as significant improvements to Siri.
Disappointing lack of details on why it is going to be the biggest. Message improvements (meh) and AI integration, that might possibly make Siri better. More to come in the following months.
I have yet to find anything that exciting with AI to care about it coming to iOS. Arc has incorporated it, but I rarely find it that helpful. My default search engine is Perplexity, but I often end up going to Google. I had my closest aha moment with it yesterday though. I needed a formula in Excel and it gave me one. It was wrong, but it was close enough that it got me on the right track. (Had I followed its source link, I would have gotten the right answer, and that source was the top Google link. Still cool though.)
I would love it if they copied Google and gave us an AI wallpaper designer (it’s only on the Pixel phones, not even the Pixel tablet, that I have).
This one cracked me up from the 9to5Mac version of the same article.
- New AI features for Apple Music to create “auto-generated playlists"
The big selling point of playlists has always been that they are human curated. But I always figured the “radio” stations, which I use a lot, were just smart playlists. Are AI generated playlists really a selling point now?
This is interesting! AI at scale is incredibly expensive. I am convinced that Apple will deploy local models into the devices it has been selling for the last 2-3 years, with cloud support for some use cases. But we’ll see.
To consider the environmental impact. CPU cycles are CPU cycles independently of wether they run in the cloud or in your phone.
Apple has always done it on device for privacy reasons, is there a way to send personal data to the cloud to process it while ensuring that Apple does not have direct (unencrypted) access to the data?
Good question. I don’t see how “the cloud” could process our data without the ability to decrypt it. So I would guess the answer is no. Apple already has first party access to everything we do on our devices so we have no option but to trust them.
I’m not sure why excitement is needed. The best AI is the AI that gets a job done better than before, ideally without anyone noticing what’s under the covers.
I have multiple applications on my Mac that leverage Machine Learning (let’s call it what it is) to very great effect. But none of the features I regularly use and rely on are generative. They just get an existing job done far better than older methods. None of these uses are ‘exciting’, though I do occasionally say “wow” at the results I get.
I think people see “AI” and think “write my paper, steal my job, break down of society” but it has been quietly improving common tasks for years, and that’s where I think Apple will employ it.
Because everything everywhere is “AI, AI, AI.” Every tech site is constantly talking about it and forums have numerous posts about it (this forum had a tiresome amount of posts on it a few months ago). Yet, every time I use it, I might find it neat, but not that useful. It’s one of those things that I bet in 10 years will be very cool, but now everyone is trying to force it into where it isn’t needed.
Take the Music AI generated playlists. Is it really any different than the current smart playlists we have now? Could it really do better than a human who is really into music? Apple has always sold it as humans can do it better, but now AI! I am extremely skeptical. It’s just more marketing because that’s the craze.
I am all for Siri getting better, and believe it could help, but Siri’s problems don’t root from how smart it is, but its implementation. I doubt AI is going to make it so I don’t have to repeat myself 5 times to turn off one light.
I don’t have to repeat myself 5 times…
Perhaps AI can get it down to only three.
Completely agree. Just because there is a lot of misunderstanding about AI or over-excitement from the tech writers out there, doesn’t mean AI can’t dramatically improve the iPhone. And I think the more impressive part will be if you don’t really notice it at all. The phone just works better, Siri just responds faster, Siri can handle more complicated tasks, etc.
I was that way until just recently I used AI in two instances that helped me:
- using ChatGPT to help me code and debug that code, and now creating problems for me to solve that it can help me with
- spitballing ideas for a logo for a project I’m working on, that I have used as inspiration when I haded it off to an artist.
So those two things to me boil down to helping me learn something and acting as a sounding board/brainstorming buddy.
And I’m sure I’m barely scratching the surface of what AI can do.
There are LLM’s out that have been run on Pentium II’s. I think it’s perfectly reasonable to expect it to run locally at first. We shall see…
Not only (a certain) speed is needed, these models are very large i.e., memory size is way beyond the capacity of a phone or local mac.
That’s true. For example, Google’s Gemini LLM comes in three sizes, Ultra, Pro, and Nano. Nano is designed to run on the Pixel 8, but it won’t be able to compete with the Ultra which runs on a data center.
Inside Apple’s Big Plan to Bring Generative AI to All Its Devices
I think there are a lot of people that would disagree. This was proven many months ago. Which feels like an eternity given the pace of change with AI…
“ ChatGPT is already the past… Run Llama 2 in your laptop for free!”
Aww, yes that is a good point. That particular article is much too optimistic (my opinion) - I didn’t consider Llama and similar standalone models as in my experience Llama feels very toy-ish (certainly for coding doesn’t come near GPT4 which already has noticeable limitations). It feels to me (speculating) these models are memory limited - hence my remark above but I should have given context.
I mean, Strunk repeated “omit needless words” three times to fill space.
FWIW, Ming-Chi Kuo is saying the big changes aren’t coming until the iPhone 17. Time will tell, I guess. iPhone shipments could drop 15% YoY in 2024
Don’t expect AI to change the world but look for ways it can improve your world.
My photography benefits immensely from “AI”.
- DxO PhotoLab uses ML in its de-noising system, which beats any non-ML option by a country mile.
- Topaz Photo AI uses ML to clean up, upscale, de-noise, and generally enhance photos. It does a fantastic job on many old photos I am scanning.
My family history benefits immensely from “AI”.
- I recorded conversations with my mother, which ML transcribed very quickly and accurately for me, to send to family members who’d rather read than listen.
I couldn’t care less about AI generating pictures or text for me. Yes, I’ve played with both, found them wanting for my purposes (including that I don’t have a purpose for creating text), and have moved on.
So maybe that’s one example where things aren’t different enough yet. But also… if no-one told you it had been switched to AI, would you have noticed or cared?
I totally get you’re sick of the hype. “AI” is just the latest subject to be over-hyped. Be like me… ignore the hype (I’m currently avoiding most Vision Pro reporting and commentary). Concentrate on what matters to you.