E.g., if I open the share menu, whatever contacts and apps Google guesses is always wrong, and a simple list of whoever I've last shared with would be better.
I also don't want any "smart" contextual behavior (widgets changing out throughout the day, etc.). The phone simply doesn't have enough context to know what I want from it at any given moment. I'd rather just tell it how to behave using my own judgement.
A decent AI system would include a heavy bias toward recent. Good "AI" should mostly feel like dumb heuristics. I think a good example is directions on my iPhone: when it's plugged into a car it always suggests driving directions. When there's something in my calendar soon it suggests going there. If there's not and I'm not at home it often suggests going home (it's right more than half the time). After that, it just offers a list of buttons for places I've told it about: home, work, school.
I'd be annoyed if it tried to do something more clever (like suggesting the post office because I went there on this day last year). I think a key element is that I have an immediate and intuitive idea of why its suggestion is a reasonable one.
It even finds text inside images.
Am sure Android has these features but I don't know if they work offline.
I believe the iPhone automatically classifies photos based on who or what show up in them. Users can contribute to train classifiers, but the iPhone already works out of the box.
The iPhone also creates theme-based photo albums from the photos you took. I recall it creating Christmas photo books, photo books featuring a pair of people, persons and pets, etc.
This might be low-key AI, but it's the useful kind.
> It even finds text inside images.
I'm not sure OCR counts as AI. But yeah, the iPhone indeed does that too. We can take a photo of a telephone number or even a credit card and automatically fill in those numbers.
I worked on a similar feature in the past for an unrelated project, but that was not AI though. Marketing changes though.
This is a great example of the AI effect where people would call something AI when it was a daunting research problem but give it another label once it’s working:
https://en.wikipedia.org/wiki/AI_effect
In this case, Apple’s modern OCR is a complex neural network system which I think most people would class as an AI tool. It’s notably better than traditional approaches which were optimized for business documents.
I'm not sure I agree, primarily because "better" is subjective.
A pipeline with template matching is extremely effective at extracting fixed form text in a standard layout, such as telephone numbers or credit cards, and computationally cheap as well.
But I presume a drop-in black box model which isn't bound to a low computational budget, can output plenty of false negatives and false positives, and can run on a single pipeline might be preferable at least from a product management point of view.
Also, neural networks looks good on resume while template matching doesn't. Just like statician/image analyst looks lukewarm but AI engineer looks superb.
Isn’t Tesseract also neural network-based?
And another toggle for 100% when I do need full charge, because I go somewhere. Why such basic functionality is not avaiable, is a bit frustrating.
Settings > Battery > Battery Health and Charging. Set "Optimize Battery Charging" toggle to off, and the adaptive charging is gone.
That way it at least serves the user, and not a paperclip maximiser for engagement.
Seems to be just another case of a half-assed Google product which is destined to die on https://killedbygoogle.com/ in a few months / years, sadly.
I'm not 100% convinced that this is by accident.
Its bad that platform that is supposed to be more free alternative is not so much so. Similar story with me and Strava - when I record some walking, Samsung's latest android 13 sometimes tends to kill after some time the tracking thread. So my maps sometimes look like I invented teleport, pretty annoying and made the app useless and unused. I checked and Strava doesn't have any settings for this enabled.
Music discovery is the best feature for me in spotify. Especially if you listen to more unpopular genres like french/german techno, old school hard trance. I found plenty of great artists there which have <5k monthly plays.
I think song/album radio is not exactly based on your taste, but on what spotify considers similar.
Try making a playlist with your favorite songs and press "enhance" or "playlist radio" to see recommendations based on that context.
On a sidenote: Looks like techno had massive growth in the last 1-2 years. I discovered most of these artists when they were much smaller. Now most of them have 3x to 10x more monthly listeners. Seems like Spotify is indeed a great platform for discoverability. For monetizing probably not though.
---
French:
KASST - Hell on Earth (the most popular on this list. 430k monthly listeners): https://www.youtube.com/watch?v=Fwi1qgSaxZY
Trudge - страсть (this dude has many decent tracks, check out "Sway over the Void" and "Self Love Club" out): https://www.youtube.com/watch?v=dx-zzugl-P0
BXTR - Reboot my soul (40k monthly listeners): https://www.youtube.com/watch?v=ruxTTzO3yTI
RAUMM - Perdue Dans Le Noir - BXTR remix (Has French lyrics. 38k monthly listeners): https://www.youtube.com/watch?v=m-xkLnJUvDQ
Automates - Meta (French lyrics too, 29k monthly listeners. One of my overall favorites): https://www.youtube.com/watch?v=Il0_Pcj4nho
Any another production one from RAUMM, which is more like a 16-minute experience with beautiful lyrics with a lot of trance influence. I'm uncertain if this still counts as techno; I don't think so. But I really love sitting down on the couch and putting this on a loud volume and just relaxing: https://www.youtube.com/watch?v=1b-KUdT5bU8
I Hate Models - Daydream (pretty famous; 500k monthly listeners): https://www.youtube.com/watch?v=xjCb4CB-hdw
Anetha ft Sugar - Candy from Strangers: https://www.youtube.com/watch?v=ofOPhZvnVF4
Trym - Millenium Pain (ナルト Mix): https://www.youtube.com/watch?v=INwOGDbQFmQ
-----
German:
In Verruf - Let Out (80k monthly listeners): https://youtu.be/pc_Pp0kObxQ?t=1
Parallx - Red Clouds (30k monthly listeners): https://www.youtube.com/watch?v=FEX5xJOuxAM
Inhalt der Nacht - Die Liebe (25k monthly listeners): https://www.youtube.com/watch?v=fVu6UBYuq80
Klangkünstler - Ihr Werdet Weinen (Pretty famous; 500k monthly listeners): https://www.youtube.com/watch?v=vFIiveXMxyI
Introversion - Dystopa (50k monthly listeners): https://youtu.be/KRXuatwFhIE?t=2
B2 - Destiny (64k monthly listeners): https://www.youtube.com/watch?v=zXaqpT323DQ
Kobosil - Rigid (Kobosil 44 Rush Mix)(Fastest growing Technio Producer in Germany, 500k monthyl listeners): https://www.youtube.com/watch?v=Uf1MLglK3fc
AKKI - The Ocean: https://www.youtube.com/watch?v=bD5GrfTy9Ac // Into Your Mind: https://www.youtube.com/watch?v=jWWAGoZmx9A
rezystor - alice (this isn't on spotify unfortunately; rezystor has <10k monthly listeners): https://www.youtube.com/watch?v=fcGa15wEalE
David Strasser - Trust the Process (17k monthly listeners): https://www.youtube.com/watch?v=yY9T6mdg2wo
----
I had to snuck a track from two Italians in here because they only have like 1k monthly listeners on Spotify and make good music:
Anticyclone - Ruins Of Las Vegas: https://www.youtube.com/watch?v=vFquUsnXKuw
I have some cron jobs running that generate new playlists every day with very specific targets (e.g. "min_instrumentalness" on a track) while still being random enough to have variety.
If Spotify ever shuts down their API like how Twitter has done, then I'll move to something else.
I went to Spotify's search and typed "i dont like' and hit enter, figuring what I was looking for would probably be near the top of the results. It was indeed the first result.
But man...the rest of the results were something else [1]. I've never done anything on Spotify that would suggest I might like most of those songs, or in many cases that I would even be interested in their genre.
Eventually curiosity got the better of me and I did listen to some of those to see if they were as bad as their titles suggested. They were.
But now I think Spotify has decided I like that kind of music.
If I do the same search now it has many of those still but now adds songs called "Sex Offender Shuffle", "Niggas in My Butthole", "3 BIG BALLS", "Alabama Nigga - Original Version", and "I Don't Like Beaners" to the top 15 or so.
There are plenty of songs with "I Don't Like" in the title or something close to which are about things like relationship problems or self esteem that are in the search results, so it is not like they only had songs about disliking certain races or songs about people's genitals to choose from.
I have had an Android phone since Froyo. I liked it when the functionality was simple and didn't try to inject too much suggestions / AI. Now, Google Photos app is trying to suggest things to print, the Settings app is trying to predict when and how to charge my battery (I just want it 80%, all the time).
The more AI Google applies to my life, the more I want to escape it. I remember, when it was said people pay a premium for Apple products because they want privacy. I wonder if it will also become, people migrate to Apple to escape AI.