That's why on my new-to-me Pixel 7a I have LineageOS 21 - already quite old - and will probably never update it. I don't keep a lot of stuff on my phone anyway because I do real work only on my laptop; so security isn't as much of a issue for me as it is for most phone users. With various programs available on F-Droid and - irony of ironies - a LOT of help from Google AI, I'm getting rid of some of Google's shitty UI and aesthetic choices. There's still stuff I can't fix, but at least the phone is now usable and (mostly) doesn't make my eyes hurt.

Even stock Android looks good in comparison to the iPhone I've been using for business. (God I hate that OS - I don't get how people put up with it, much less like it). When I get my personal phone straightened away fully I'll get another Pixel and repeat the LineageOS dance.

If you're going to put power in the hands of users, go all the way and give them full control. This bullshit of treating some websites as sacrosanct works of art by full-of-themselves web developers needs to end.

Once it's on my computer I should be able to do whatever the hell I want with it, short of violating copyright laws.

I still try to use AI-free DDG fairly frequently, and it seldom works anymore. Its results are so bad as to be unusable - so I'm pretty sure the AI is totally absent and not "hidden" or "obfuscated". If AI was in the mix at all, the results would be better.

The dystopia was here long before AI became pervasive. AI just makes it both worse and more obvious.

I often have multiple programs partially overlapping, and dragging the cursor from side to side raises the middle app(s) when that's not where I'm going. I find that distracting and annoying.

It may be my ADHD. There are several other newer UI features that I detest and spend stupid amounts of time figuring out how to squelch. "Smooth" scrolling? To me it's just the opposite - I call it jerky scrolling, and it drives me nuts. I kill that fucker wherever I can.

Then there are scrollbars that are pencil-line-thin until the cursor hits them; at which point they widen to Sharpie-line thin, which is still too narrow to be usable. Give me a scrollbar wide enough to grab, with steppers, then leave it the fuck alone so I don't have to play whac-a-mole to use it. Ditto for window controls.

I have nine different text files describing how to get the UI (mostly) the way I want it, in a folder called Fix_UI_Fuckery. Not that I'm fussy or anything...

Good point. But I'm betting that a human CEO could out-hallucinate the silicon variety all day every day...

Even just for archiving, I would have thought tape would be too slow for the volume of data involved in AI farms. And don't tapes still occupy a lot more space than equivalent-capacity hard drives? Also, what about access times? I know it's just archiving, but still...

I fail to understand how AI can flatten management structures. I can see the argument for reduced head count, but I would have thought relying on AI for mission-critical work would require more oversight and review and testing, and therefore more levels of management. Which might, of course, largely or entirely offset the saving associated with reduced head count. For that reason, I also find their "less overhead" claim to be a bit questionable.

Then there's the cost of "compute", as the cool kids call it. Is SUSE going to pay for hardware and management of its own AI servers and models? Or will it outsource some or all of it? Either way there's a cost involved, and it would be interesting to see their estimates of both that cost, and the saving from "rightsizing" or whatever other euphemism they might use.

Found the parasite! Its presence probably explains a lot.

On the plus side, in looking for more info I discovered that BOFH is still a thing: https://www.theregister.com/ta...

I'd ask why they think just changing the name will fool anybody. Sadly though, mere name-changing has proven its ability to fool a huge segment of the population.

It was mostly name-changing of various kinds that led to today's extravagant corporate excess and decadence. And it was mostly name-changing that led the electorate, purposely dumbed down by under-education and propaganda, to put the current US administration in power.

Part of me wonders if this is yet another example of Microsoft currying favour with President Bone-spurs. I'm sure that inconveniencing us "nasty" Canadians earned dear Satya a few brownie points.

I don't think anybody can help a company whose approach to installing a light-bulb is to hold the bulb in place and rotate the building. Oh, and when they fail at that they simply declare darkness as the new standard.

I would think that an audit trail only needs temperature readings, not adjustments.

Thanks for the insight.

As for ...

... Did you have it code individual modules, then vet each carefully before moving on to the next building block? I'm guessing that approach might be what differentiates you from 'vibe coders' who give the LLM specs for a whole project, let it run, and take the results as a finished product.

I'm starting to think that crappy AI code is just like crappy pre-AI code: it results from a lack of knowledge, diligence, proper testing, and a proper spec against which to compare the resulting application. Does that sound about right?

From the POV of a non-programmer, I'm wondering if using AI effectively might be at least as much about having it answer questions as about writing the code. Note that those questions may include having one AI check another AI's homework.

Just recently I've been using AI a lot for the programming-adjacent project of putting LineageOS on a Pixel phone. I was using ChatGPT, and some of its advice didn't smell right so I decided to search for primary info beyond what was on the LineageOS site. I tried DDG, but its results sucked, possibly because I have the AI feature turned off. So I went to Google, where AI is pretty much unavoidable.

Google contradicted ChatGPT, and its answers seemed more in line with my past and present experience. It also assumed that I was asking for instructions. The ones it gave were clear and seemed quite sane. So I ditched ChatGPT and ended up using GoogleAI to guide me through the process of installing the OS and dialing in the system to my liking. It took some wrangling and a few trips down dead-end alleys, but overall I've been very happy with the results.

I can easily imagine that in other cases, ChatGPT's output would be better than Google's. So maybe using one AI to check another AI's output would also be worthwhile when doing real programming?

So far I'm thinking of it as a zero-sum game. One side makes gains in exploiting vulnerabilities, while the other gets better at finding and closing them.

But then again there's the unknown of using AI tools to write code. (Unknown because it's still early days). Will that result in a lot of buggy, vulnerable, half-baked apps just begging to be exploited by AI and non-AI alike? The returns aren't all in, but results so far make me think we may be in for a world of hurt on that front.