Off-and-on trying out an account over at @tal@oleo.cafe due to scraping bots bogging down lemmy.today to the point of near-unusability.

  • 92 Posts
  • 5.39K Comments
Joined 3 years ago
cake
Cake day: October 4th, 2023

help-circle
  • The research firm purchased every subscription from the two AI providers and discovered that the approximate maximum possible spend (assuming API pricing) is far larger than what users pay every month. For example, Claude Max 20x costs $200 a month, but maximizing it would cost $8,000 a month in token spend, while ChatGPT Pro 20x, which is also $200 monthly, has a maximum possible spend of around $14,000.

    Ehhh…yeah, but that alone isn’t necessarily an issue. There are plenty of services that exist that rely on consumers, in aggregate, not maximizing resource usage. Residential ISPs normally oversell their service. That works because the typical user only uses a tiny fraction of their sustained maximum rate of bandwidth consumption. In theory, if a lot of users started fully saturating their lines all the time, ISPs could shift everyone to metered service, but it works well enough and enough people value not having to worry about metering more than paying the minimum per-byte cost, so the system functions.




  • As long as you have paging space — and you almost certainly do — in general, an OS should page out pages of memory used by software that isn’t being frequently used.

    It’s possible that the program is allocating memory and then keeps hammering it to keep it resident, but most of the time, a program allocating more memory than it absolutely requires isn’t honestly that big a deal, because the OS will handle it reasonably and just leave it in the pagefile.

    I don’t know what the mechanism is to report this on Windows, but you might try looking in mmc (Windows-R mmc Enter), if that’s still a thing in Windows 11, as it has a bunch of process resource usage graphs.

    On Linux, you’ll probably want to be looking at RSS to find what’s actually in physical memory (well, there can be shared memory and some other things, but RSS is probably a reasonably approximation). You can see this in top.

    searches

    If this is current, Windows probably calls it “Working Set”.




  • A federal judge in Mississippi has punished all four lawyers on opposing sides in a civil trial and canceled the proceedings after some of them, relying on artificial intelligence, cited fake legal cases in court filings.

    https://www.nytimes.com/2026/06/09/us/ai-lawyers-sanctioned-mississippi.html

    In an order filed on Monday, Sharion Aycock, a senior U.S. District Court judge, wrote that the four lawyers had violated Rule 11 of the Federal Rules of Civil Procedure when they certified that the information in their filings was factual.

    I think one concerning thing is that this is the easiest thing to check. I mean, at some point, I assume that someone is going to rig something up to LexisNexis to actually validate the existence of cited cases, because that’s pretty simple and mechanical. Heck, even those lawyers, even if they don’t have any tech people at their fingertips, could have had a paralegal check citations or something. It really shouldn’t be that fundamentally hard for a lawyer to avoid getting in trouble for this specific issue, even if they generated the text with an LLM.

    My bigger concern is that if lawyers are willing to put stuff like this out, they’re presumably also willing to put out information that hasn’t been checked where the errors are subtler and it’s harder to find erroneous material. In the case of citing nonexistent cases, it’s really easy to say “the lawyer clearly didn’t even look at this”, because it’s hard to make that kind of error if you have read over it. This is, once highlighted, flagrant and obvious. But…there’s potential for subtler errors, where it’s harder to tell whether the lawyer did at least try to review the material and just made a basic error, and thus it’s harder to impose punishments for it.


    1. Lamplighter

    Lamplighters were responsible for lighting and extinguishing gas street lamps in towns and cities before electric lighting became standard. They typically carried ladders and torches to perform their duties. The job was crucial for maintaining public safety during the evenings. However, with the introduction of electric streetlights, the need for manual lamp maintenance disappeared, leading to the decline of this occupation. Lamplighters are now part of history, representing a bygone era of urban infrastructure.

    The lamplighters themselves were machine operators that replaced earlier professions.

    https://en.wikipedia.org/wiki/Link-boy

    A link-boy (or link boy or linkboy) was a boy who carried a flaming torch to light the way for pedestrians at night. Linkboys were common in London in the days before the introduction of gas lighting in the early to mid 19th century.


  • https://en.wikipedia.org/wiki/Washerwoman

    A washerwoman or laundress was a person, usually a woman, employed to wash laundry by hand, before the widespread use of washing machines and commercial laundries. The profession existed in many cultures, spanning from antiquity to the early modern period. While the profession has historically been gendered, often associated with women, in some contexts, men also performed laundry labor. It was typically low-paid, physically arduous, and associated with lower social status.

    The occupation began to decline with the rise of commercial laundries. The spread of domestic washing machines and self-service laundries further reduced the need for the independent washerwomen profession. By the late twentieth century, the profession had largely disappeared in industrialized countries.




  • https://en.wikipedia.org/wiki/Filling_station_attendant

    A filling station attendant or gas station attendant (also known as a gas jockey in the US and Canada[1][2]) is a worker at a full-service filling station who performs services other than accepting payment. Tasks usually include pumping fuel, cleaning windshields, and checking vehicle oil levels. Prior to the introduction of self-starting vehicle engines, attendants would also start vehicle engines by manually turning the crankshaft with a hand crank.

    In the United States, gas jockeys were often tipped for their services,[3] but this is now rare as full-service stations are uncommon except in New Jersey, 16 “urban” counties in Oregon, 4 cities in Massachusetts, and the town of Huntington, New York, where there are laws or restrictions against letting customers pump their own gasoline.




  • tal@lemmy.todaytoAsk Lemmy@lemmy.world...........
    link
    fedilink
    English
    arrow-up
    3
    ·
    7 days ago

    checks

    https://lemmyverse.net/communities has a list.

    It’s on the lemmy.porn instance, !forcedincest@lemmy.porn. Presumably that instance is, well, for porn.

    If you don’t want to see it, you can block that community and you won’t see any more from it. If you don’t want to see anything from that instance, you can block that instance. If you don’t want to see NSFW content in general, you can block that.

    As to “why” in the broader sense, it’s because the Threadiverse is a global system. It spans many countries and different groups of people. It’s like asking why something is “allowed” on the Web — everything is allowed as long as the local country is okay with it.



  • Magewell Pro Capture card

    I’ve been kind of shifting towards use of USB devices over internal cards.

    All of the USB devices that I have still can be connected to computers. Ditto for DE-9 serial ports, though I might need a USB adapter.

    But I’ve seen ISA->PCI/AGP->PCIe obsolete a lot of old hardware that I’ve had sitting around, and that’s just on the PC. That includes my video capture hardware.




  • “use a strong password” whats that gonna do if the database gets pwned, sandra?

    Strong passwords aren’t intended to simply protect against brute-forcing a password via trying to authenticate repeatedly, but also to help protect against brute-force attempts to obtain passwords from a compromised password database using a dictionary attack, the scenario you’re describing.

    Typically — if an authentication system is storing its password database competently — the password shouldn’t be stored in plain text. Instead, the password will be salted (to avoid rainbow table attacks) and then hashed via a cryptographic hash. The password database entry will look something like a tuple of (username, salt, salted hashed password). If the password is a strong one, it will be computationally-hard to obtain the plaintext password, even if someone has the salt and the salted, hashed password.