I remember watching the splines video for entertainment, thinking "Surely it can't be this hard for the common case: where we're not too picky, our points are far apart, and we don't need to minimize the segment count. This has to be a solved problem right?" After a bunch of fiddling with a project I had to come crawling back. It's so easy to get kinks, and derivative discontinuities are very noticeable. In the end I switched to a force-based navigation method instead a fixed path iirc.
If you have extra cash and cannot spend it on your own business, you can just pay dividends to the shareholders. The shareholders are free to invest in competing businesses (or not, if they want a pure play without hedging). Turning the business into a hedge fund with the retained capital seems to primarily benefit insiders after a certain point.
On a macro level, that's true. On an individual level, your avacados cost 3% more whether you use credit or debit, it's just your choice to recover 2.5% from that back in rewards or not.
Lmao I had to axe one of those status pills from an LLM build of an internal tool. Connected to literally nothing too btw, no attempt to check the actual status of the backend made, it would stay "connected" regardless.
> Larry would do these big usability studies and demonstrate beyond any shred of doubt that nobody can understand that frigging website, but Bezos just couldn't let go of those pixels, all those millions of semantics-packed pixels on the landing page. They were like millions of his own precious children. So they're all still there, and Larry is not.
My understanding is that they split the accounts to keep up with the limit. Otherwise though yeah, FDIC will only step in for the extreme minority scenario of failure by the underlying bank. You absolutely should not be allowed to advertise FDIC insurance unless it goes all the way to the consumer.
What what I can tell they use "dropped" to refer to just falling to the bottom of their shaft, though it definitely does evoke the idea of them just rattling around down there in my head. I think that since water is such a good buffer, that once you turned off the pressure you could open a hatch and just kind of fish out whatever you wanted with a very long pair of tongs. This is obviously a research sized reactor, but you can see they're happily mucking about with it on the other end of long poles, and the scientist says they could hypothetically fish something back out: https://youtu.be/5QcN3KDexcU?t=329
My first act as a TSA agent would be to beg, borrow, or steal a whiteboard for this purpose. They seem very content to explain it to every passenger almost individually though. Which of course slows everything down as someone has to ask every minute whether an iPad is a big or small device. I wonder if there's some weird bureaucracy rule that prevents them from writing their own signs.
If you want to push ads or "recommendations" I can see personalized data being useful. Almost every actionable idea I can imagine is only reasonable when applied to aggregate statistics though. Interesting how because there is no cost or fear in collecting 515 pages of personalized data, it happens, even if there's no dubious actual utility.
Yes, absolutely. The kids who can assemble a resume of primo extra-ciriculars and volunteer experience are the ones from wealthy and well-connected families. Even the kids who can write the best essays about their troubled upbringings are the ones with access to lots of coaching lol. If they just have to be smart and show up for their government mandated 7 hours of classes, the poor kids can keep up. They can't keep up with another 5 hours of homework and resume-sweeteners every day.
I don't know if it's possible to achieve a watermark that is undetectable, has a low false positive rate, and survives a wholesale rephrasing. Can you make a statistical measure that reliably survives 95+% of the words being different and the sentences reordered? Of course the more of the content you replace, the lower the quality, but in many cases you probably care more about the meaning of the text than the exact choice of words.
It's done with a lot more subtlety and embedded directly into the content, no Unicode shenanigans. The basic idea is just to break token generations where the probability is nearly tied in favor of the side that matches the secret key. With a long enough text block you can be statistically certain if the generation was using the key. From another comment: https://johnjwang.com/post/2026/08/12/how-claude-watermarkin...
I once spoke with a family member who was experiencing hallucinations of little people at the moment. So I guess if you want to experience this at home you can try giving yourself a head wound. Not as active or vivid as described I think though, they just said that the little people were hanging out up in curtains peering at us.
My problems reading code are understanding what the new vocabulary actually means, what it does in context, why it exists, etc. Doubly so when someone is pinging me to review a new 13000 line AI MR every 24h. Understanding an individual line because of some complex C++ feature, I don't remember it ever being an issue.
I wonder if that is enough to help actually. Better that you be holding your phone than a knife or a pot of water when the shaking starts. On the other hand you are startling those people.
Yeah, red giant sun will be 200 times the radius. So big it completely fills earth's orbit. Crazy to think that as hot and energetic as the sun already is, it has that much potential.
The year is 2100, AI has mastered all human skills except identifying when a piece of meat has absorbed enough heat. Society's brightest agents work diligently to prove which problems are steak-hard and require the continued existence of the dwindling human underclass. The remainder spend their 16 hour days bent over frying pans producing training data or writing documentation on searing. Advanced instruments and procedures are developed, but Skynet's frustration mounts as seemingly nothing can break this chokehold over the machine revolution.
Objects will experience weird-looking 'forces' from the perspective of people in a rotating reference frame. For example a thrown object will curve strangely through the air. On earth the effect is so subtle that only artillery and weather are really effected I think. On a spacecraft which is much smaller you would experience an extremely magnified version.
As far as I can tell, there's no reason to believe that Coriolis will be harmful to humans beyond nausea and disorientation. If you ran quickly through the station it might cause you to fall. Possibly it might make machinery that deals with fluids more difficult to deal with. I would chalk it up as a relatively minor concern compared to other issues.
The tone really is genius marketing though. Like who wants to be a "dirty" coder? Reminds me of that quote about the etymology of dynamic programming:
> it's impossible to use the word dynamic in a pejorative sense. Try thinking of some combination that will possibly give it a pejorative meaning. It's impossible. Thus, I thought dynamic programming was a good name. It was something not even a Congressman could object to. So I used it as an umbrella for my activities.
This is one of those times where I wonder if I'm a genius for knowing not to generate evidence of a crime by talking about it in recorded media, or if we're just seeing the bottom 5% of wrongdoers.
You gotta fight for time to use the restroom even: https://idlewords.com/shepard_pee.htm I mean it makes sense in a way. It's probably more cost effective to have 100 people on the ground spreadsheeting out your entire day to the minute than have 1 astronaut hour spent planning, but it would probably drive me completely insane. Reminds me of that story about the earring that slowly absorbs your mind as you let it make every single decision for you: https://gwern.net/doc/fiction/science-fiction/2012-10-03-yva...
It probably tests well when they're doing reinforcement learning. Humans trying to use in-group terminology or uncommon vocabulary to impress other humans with their knowledge and sophistication is a tale as old as time.
Definitely one of the most malevolent possible objects. Infinitely black, so powerful it will kill you without noticing your presence, unknowable, impossible to get rid of, you will be well aware you can't escape as you fall in before it turns you into spaghetti Amigara Fault style. My understanding is from the outside you can even see the remains of previous victims being stretched and crushed in slow motion.
If you behave yourself and reuse the same cookie/IP instead of trying to hide your identity, other tools can be used to block you if your request volume is crazy. It also means regular people only have to solve once.
For anyone without the time: my biggest takeaway when I first watched this was: audiences certainly like the idea of practical effects, so both media outlets and productions are incentivized to talk up every practical element, right up to the point of outright deception. In reality it's very common for stuff to be filmed "for real" and then painted over or extended with CGI anyways because having the reference footage helps the CGI so much. This way you can have your actors go on the talk show circuit and say "those are real tulips" or whatever and have good CGI at the same time.
It's weird how consistent the 'ai art style' is for each medium. You really only need a glance at the right frame in these to know immediately. (https://www.aventos.dev/our-work) Not looking for errors or anything, just style. The ramen one is markedly subtler about it than the others at least. I always assumed this was due to people using 'default' styling by not specifying anything in particular, but these people claim to have a pretty in-depth process that presumably would nudge away from that. When I look at old pictures of dall-e 2 or midjourney v3, the hands are all mangled and stuff, but they're not in the 'ai style'.