Do you have a guide on which baseline rules are good? I use Claude Opus 5.5 at work and while it’s definitely more helpful than not having it, it also does some incredibly baffling things that a human would realize immediately are unproductive.
This week, because I’m wiping old data for someone who wants to get rid of computer equipment.
In my experience, the LLM usually proposes a solution, but it’s a crapshoot whether it works well. And plenty of bugfixes go well beyond a few lines, especially in older code.
It was originally conceived to replace Android, but then it got relegated to Google Home devices, and now it’s basically just a REST client for Gemini that’s bleeding users because Gemini on smart home stuff sucks even more than Gemini normally.
Check out the Fairbuds 2, then. The batteries still degrade, obviously, but you can reasonably replace them.
I’m starting to think that Carl Pei gets off on making good but somewhat unfinished phones for a few generations and then following them up with bloated, confusing garbage. OnePlus did the same thing.
They’re much closer than e.g. Samsung OneUI, though.
And Fuchsia! RIP Fuchsia.
Sideloading has been a word for installing apps that aren’t from a store at least since I started using Android in early 2010. There was never any negative connotation, aside from the acknowledgment that you wouldn’t get automatic updates.
“Cleaning up” seems a little generous. It’s finding them, but someone still has to patch them.
I’ll probably pirate Ocarina when the Switch 2 gets CFW and buy GTA when it hits $15ish.
Artificial superintelligence is a specific thing that doesn’t exist yet. A ban on it would be a cautionary measure in case of future advancements. A ban on artificial intelligence writ large would arguably ban my high school project that played Tic-Tac-Toe optimally.
It’s probably important to note that ASI (and its more conservative form, artificial general intelligence or AGI) are goals, which is to say that they describe the result and not the technology used to get there. LLMs are a specific technology that I’m pretty convinced will never achieve AGI, let alone ASI. Specific-purpose AI has been around for decades (e.g. Deep Blue, which was world-class at chess in the '90s), and LLMs aren’t really any different; their specific purpose is completing sentences, though, and that tricks a lot of people into thinking that they can actually perform metacognition in the way that an AGI would.
Basically, if you presume freedom of will and peraonal responsibility are a thing then it is significantly more likely that there is a possibility for a person to lift themselves out of poverty one way or another and they choose not to act upon it (rather than strictly corcumstances that make it impossible to do so).
That doesn’t follow. You can have lots of bad options and the freedom to choose between them.
That is not to mention that the wast majority of the people are simply bad with money. Like they would take out a loan on a new iphone when they can eaaily get a cheap paidoff android/a used iphone or they would buy a car that is way above whet they can comfortably afford
Do you have any data that suggests that this is in fact the vast majority of people? There are certainly people who do this, but there are also people who choose a newer car because they can’t afford the uncertainty of potential maintenance problems with an old one, for example. The newer car costs more on average, but it’s more consistent. When you don’t have savings to smooth out the bumps, you could lose a lot if a problem happens at a particularly inconvenient time.
A phone is such a negligible expense compared to a car that I don’t think it really needs addressing. Even a maxed-oit iPhone Duo is $100/month or so. That’s less than most people spend on gas alone.
And that’s ignoring people who get locked into payments on things that they could afford when they bought them, but no longer can. Job loss or medical expenses can wipe out your savings in a hurry.
Ok, IRA is a strictly amercan thing (I guess)
The cooler version is strictly an Irish thing.
Im afraid Its not that the capitalism game is rigged,
It is.
its just that you are bad at playing it.
It can also be that, but it’s volatile enough that no strategy is really optimal and even the better strategies can leave you in the hole.
(which is putting money YOU WONT NEED in another 10+ years in there EACH MONTH and not even lookimg at it for the bare minimum of ~10y).
Some people don’t have money that they don’t need each month. Most people don’t have a guarantee that they won’t need the money for 10 years. Remember, we’re talking about the US. The social safety net here is made of cobwebs. That $10k you put away over the last three years might suddenly be $8k that makes the difference between staying afloat and losing your housing.
Also, even ignoring taxes, it’s pretty common for tech products to cost the same number of USD, EUR, and GBP.
Here are some examples. Keep in mind that this is just companies that have rehired and admitted it; others are rehiring under another name (e.g. “we cut our workforce because of AI, then it was so successful that we grew and added more jobs” when they’re actually the exact same jobs) and still others are just taking the traditional path of laying off people who were actually needed and then shoving the work onto the remaining employees, degrading their products, and so on.
Also, a lot of companies labeled their layoffs as AI not because they actually had anything to do with AI, but because the company was in financial trouble and needed to lay people off, but claiming it was due to AI made the stock price go up instead of down.
A bunch of companies are panicking and rehiring people because the LLMs are worse, slower, and/or more expensive than people, though. Causing people to lose their jobs doesn’t require it to actually do the jobs better, it just requires CEOs to think that it can replace people, which turned out to be far easier.
Tier 1 companies also lay people off pretty frequently, which is highly inconvenient if you rely on them for a visa, and that’s assuming you get a job there in the first place. I’m not going to say it’s never a good idea to try for that, but in this environment, your effort is probably better spent elsewhere.
A 95% false positive rate isn’t actually that big a deal if you have a good way to test the claims. I wonder if you could set it up CtF-style, where you don’t get notified until the LLM can write to a specific file on the target box.