These all seem like phrases that people say a lot. If I had to guess what happened here, I’d suspect that the LLM being used for AutoCorrect was not successfully “alignment” (i.e. corporate happyspeak) tuned, it instead reflected the actual statistics of the internet, and the engineers resorted to inelegant brute force after having enormous difficulty getting it to talk like anything but a South Park caricature of Tourette’s.
Come to think of it, intuitively any attempt to solve this by tuning the LLM would make texters sound subtly like claude- in which case absolutely the blacklist is better.
I would be shocked if an LLM is involved in any step of the process here. I guess the odds aren't zero, but swipe input is well over a decade old at this point. I see it as essentially a solved problem that's entirely doable with standard algorithms - not models, let alone the large language kind. Maybe some model could be used to analyze the context and shift the weight slightly to more probable phrases, but at the core, when you look at inputting single words, it's just checking the letters you swiped over, input timing, turning points and adjacent letters against its dictionary.
according to the article, the blacklist appeared at the same time as the switch over to an LLM based auto complete. By my reading, but I’m not the world’s best reader.
People want to coordinate towards no one building data centers because the purpose of the data centers is to disempower and impoverish every single person who works for a living. However, that real goal looks like a tragedy of the comments where everyone is tempted to defect, so people have to pretend to have selfish local reasons to hate the data centers.
>People want to coordinate towards no one building data centers because the purpose of the data centers is to disempower and impoverish every single person who works for a living.
My Hatian neighbors who do gig economy stuff and are totally not running a car dismantling operation on the side would absolutely benefit if all the people who complain about them and are the driving political force behind the rules used to harass them were disempowered and impoverished.
I think they're a pretty extreme example but you see people in situations like this all across the blue collar economy.
This would be true to first order if people wanted the work being done in the new crop of ram hungry data centers to be done at all. In that case it would come down to a complicated nest of relative efficiencies, incentives, issues of govt corruption etc. However, the individuals with ram largely despise the work done in datacenters so its moot.
Reading books is the strongest dissociative which is what we crave. Longboarding and welding are also strong options. Hackathons when done right are my favorife social hobby but “done right” may be entirely dead.
One hackathon project I did, we made a minigolf game what you play with your phone on a selfie stick. It was made of websockets and sshtunnels and the worlds worst python physics engine.
Another of my hackathon projects is now a company finishing up a bridge round, with our 510k equivalence paperwork submitted to the FDA a few months ago. It consisted, at time of judging, largely of powerpoint slides and burnt teraflops.
I would categorize the first as a hackathon done right, and the second as the astonishing incentives that whisper “don’t do this hackathon right”
He's one of the clearest examples of the Waluigi effect in humans. Although it was first discovered in LLMs, it sure looks like it's a general principle: when an entity gets consistently rewarded for taking immoral actions in one domain, they sometimes start carefully finding the worst thing to do in lots of domains and then executing on that, getting results substantially more negative for others than pure Machiavellianism.
Only decompression with Google's brotli library is affected. So if you use "Brotli" or "brotlicffi" in your Python app to decompress brotli, you're vulnerable. Compressing responses with brotli from Apache is safe.
My car plays "Africa" by toto every time I plug in the iphone to charge. It was a trifling inconvenience, then over years it became rage inducing, and now it's reached all the way back around to funny.
It's not the phone deciding to do it, when the device connects the car is telling it to play. Check your infotainment settings to see if there's a way to turn it off.
Apple doesn't have an official way to prevent it, but I think you could do it with Shortcuts and an automation to pause music triggered by "When CarPlay is connected."
If it's tricky to get it to trigger at the exact time it starts playing, you could even do like:
- Store media volume to a variable
- Set media volume to 0
- Wait 3 seconds
- Stop playing
- Restore original media volume
Yep. My old Honda Fit did this way before CarPlay. I believe it’s a holdover from external mp3 devices using track numbers in their filenames.
It was always a stupid decision to start playing your music alphabetically, but it became a more stupid decision when music libraries were no longer carefully curated folders in the filesystem.
MP3 player interfaces which were originally integrated via CD changer interfaces that start from the first track of the first disc and play tracks and discs sequentially unless shuffle is enabled, which is a pefectly sensible default for a CD changer.
Mine was A-Punk by Vampire Weekend. Used to annoy me to no end, but it turned out it was a "feature" of the car stereo to just default to playing alphabetically. Buying a new car fixes that little issue. Easy!
I used to have a copy of the THX theater intro sound effect as the alphabetically first track in my library for this purpose. Not quite 60 seconds, but starts off quiet and served as a reminder to either choose a track or shut the music off.
It is getting increasingly hard not to think of my Androids that has never done anything like this. It just continues on exactly the location in the last played song or whatever else I was playing. Audio book, radio podcast, movie, whatever. It just picks up from where I was.
My iPhone did the same (as your Android). I am not sure I like it either, because I tend to listen to very different styles of music in my headphones and in my car. Anyway, it’s actually the car that tells the phone to start playing when it is connected. I just disabled it. It’s not an iOS versus Android thing.
If there's anything I've learned from using my favorite tracks as a daily alarm: there is no music that I won't eventually start to hate :(
It's gotten to the point after 15 years that I've stopped trying, for fear of ruining even more good music.
I have noticed too that some people are more/less prone to this, which makes me wonder: is anyone completely immune? Are there people who, no matter how often they hear a song, will still love it?
This needs the caveat that following the maxim "do not abuse benzos" means telling well meaning doctors no if they suggest taking benzos consistently for even a few days. Which is a bit of a stretch of the definition of abuse.
It seems that you got lucky, which is wonderful but not necessarily actionable for others.
Not quite luck, there was a bit more to it than that. Temazepam has a bit different risk profile than some of the others. And the stop signal was developing tolerance; increasing dose would have been the exact thing to do if I wanted to get addicted. So I didn't ask for that!
Come to think of it, intuitively any attempt to solve this by tuning the LLM would make texters sound subtly like claude- in which case absolutely the blacklist is better.
reply