Thoughts on Android / Samsung After A Year

June 9, 2024

Apologies

It's been a long time since I wrote on this blag. Getting close to a year in fact. It is however, a bit fortunate in some ways that my last info-dump, was a long term overview on Apple's mobile platform from the point of view of someone steeped in the FLOSS software ecosystem. As I want to very much revisit my conclusions from that post.

Prior Conclusions

I had stated previously that things had been increasingly rocky for most, if not-all android forks; and that iOS was structurally bias against floss staples such as NextCloud and KDE Connect. Which left me feeling like a proprietary Samsung phone with most Google Apps disabled as my best option. Now coming in at close to a year later, I recant.

Issues With Samsung

It's tempting to see Samsung's services as a viable alternative to Google's. They very much can be. Samsung Pay is a seamless experience facilitating contactless payments everywhere in the United States where you can also use Google Pay for example.

However, in practice, this leads to a lot of duplication of features. With a Google Play Store barking for app updates, as well as a Samsung app store. Often times, the Samsung equivalent of services feels like a tedious extra superfluous layer tacked onto the operating system begging for one's attention.

Samsung's Galaxy Store is not full-featured enough for one to wholly rely on it, yet requires one to participate in to keep all of one's apps up to date. So disabling play for the Galaxy Store is also not functionally an option. This is bizarrely the expected experience.

I have also experienced Samsung updates, which turn on; and hid the ability to turn off Samsung marketing notifications for partner apps. Which is frankly disgustingly user hostile.

LLMs and Android

Google driven Android has always had privacy concerns and has been historically hungry for user data. However, Google has been chasing the AI bubble as terrifyingly quickly as Microsoft. With Google training its Gemini/Bard AI on user's text messages in Google Messages, and toying with letting it's AI to listen in on one's phone calls.

Samsung if also launching their own suite of AI features OS wide under the Galaxy AI moniker.

Now admittedly, some of these (but not all) are opt-in, and on-device. Many of them even have genuinely useful purposes. However, we haven't had a historically great track record of both organizations not employing dark patterns, or being wholly honest with respecting user data.

The Generative AI Craze Is Not Consumer Driven

I think it's clear to any industry observer that generative-AI is being viewed as a possible new capitalist expanse. Everyone is chasing it in a massive land rush, as not to be late to the game and cut out. As Microsoft was in the early days of the mobile phone wars.

I genuinely believe this AI features we are seeing, are not landing because of user-demand. The vast majority of companies, Google and Samsung included, are clearly throwing everything they can at the wall to see what sticks. If they fulfill a genuinely useful use case for somebody and is done in a privacy respecting way. It is quite frankly a happy accident. These features are not driven by user demand or need; and making them cohere to anyone's privacy strategy (as much as they can) is largely an afterthought. The “gamble” that generative AI is the next new thing, and fear of loosing one's position, is driving the jockeying in the market.

I don't plan to expound on that. This is the premise I proceed from. If you break from me on this; then we disagree on fundamental premises.

My Experience With It

I attempted to try out various AI art generators last year, to generate art for a role-playing game, I am currently running for a group of friends; and generally speaking I found pretty much every model lacking compared to the scene I was imagining.

I've also played with ChatGPT a handful of times out of morbid curiosity; testing “anti AI” strategies in online learning environments a handful of times.

I've otherwise not touched the thing, except for AI image upscaling, which I find useful for my day job.

If you are curious, Upscayl is a pretty fabulous FLOSS privacy respecting offline AI image upscaler.

My Reluctance To Use AI

My experiences with it are purposefully narrow. I quite frankly am not interested in even my input prompts into such bits of software, being consumed and digested to make the products better. I frankly view this as labor theft.

Conclusions

What I have come to conclude, is that I'm incredibly uninterested in machine learning; and sick of the whole thing. I do not care if it's useful. I don't much see the everyday value in the vast majority of the applications it is being pitched for, and I am deeply concerned about the potential environmental impacts of generative AI, it's privacy implications and the cultural impact of its hallucinations on a culture which struggles with information literacy as is.

I desperately want to opt out of it being integrated into everything. If I use it. In so far as it exists as a technology I use, I only want to have specific dedicated tools, that allow me to use it in very specific contexts, preferably offline.

Luckily, in terms of my PC, this is pretty easy to do for me. My Linux distro of choice makes it frankly a non-issue. Where I don't have to worry there. However, this means I need to leave traditional corporate controlled Android. As the AI hype machine is going to get a lot worse, before the bubble finally pops.