I am the journeyer from the valley of the dead Sega consoles. With the blessings of Sega Saturn, the gaming system of destruction, I am the Scout of Silence… Sailor Saturn.

  • 5 Posts
  • 391 Comments
Joined 1 year ago
cake
Cake day: June 29th, 2023

help-circle


  • Here are the results of these three models against Stockfish—a standard chess AI—on level 1, with a maximum of 0.01 seconds to make each move

    I’m not a Chess person or familiar with Stockfish so take this with a grain of salt, but I found a few interesting things perusing the code / docs which I think makes useful context.

    Skill Level

    I assume “level” refers to Stockfish’s Skill Level option.

    If I mathed right, Stockfish roughly estimates Skill Level 1 to be around 1445 ELO (source). However it says “This Elo rating has been calibrated at a time control of 60s+0.6s” so it may be significantly lower here.

    Skill Level affects the search depth (appears to use depth of 1 at Skill Level 1). It also enables MultiPV 4 to compute the four best principle variations and randomly pick from them (more randomly at lower skill levels).

    Move Time & Hardware

    This is all independent of move time. This author used a move time of 10 milliseconds (for stockfish, no mention on how much time the LLMs got). … or at least they did if they accounted for the “Move Overhead” option defaulting to 10 milliseconds. If they left that at it’s default then 10ms - 10ms = 0ms so 🤷‍♀️.

    There is also no information about the hardware or number of threads they ran this one, which I feel is important information.

    Evaluation Function

    After the game was over, I calculated the score after each turn in “centipawns” where a pawn is worth 100 points, and ±1500 indicates a win or loss.

    Stockfish’s FAQ mentions that they have gone beyond centipawns for evaluating positions, because it’s strong enough that material advantage is much less relevant than it used to be. I assume it doesn’t really matter at level 1 with ~0 seconds to produce moves though.

    Still since the author has Stockfish handy anyway, it’d be interesting to use it in it’s not handicapped form to evaluate who won.










  • The couple’s long conversations about transgender people haven’t gotten as much attention, with just 2,500 views on a video in which they “explore the fine line between genuine gender dysphoria and the allure of a ‘trans cult’ that may lead non-trans individuals to make life-altering decisions.”

    Oh boy an hour long video about how the trans are transing eachother as part of a trans cult!

    But why aren’t they concerned about kids being tricked into the life altering decision of being an “anti-woke” podcaster? 🤔

    Edit – made it through most of that video (how??) and it’s a bunch of reaching and fear mongering while trying to sound like principled conservatives. The funniest part was probably them blaming anime avatars for trans people.

    Edit – just kidding the funniest part was the guy calling trans people the “priest class” of urbanism. Worship me and I shall bless you with seperated bicycle lanes and bountiful farmers markets!


  • Yeah absolutely. This is happening right on the coattails of that Character.AI suicide too so it’s not like a freak impossible to predict accident. I mainly posted it because it flies in the face of all the talk of AI safety and “responsible AI practices”.

    Like Google says in their AI principles:

    We will continue to develop and apply strong safety and security practices to avoid unintended results that create risks of harm. We will design our AI systems to be appropriately cautious, and seek to develop them in accordance with best practices in AI safety research. In appropriate cases, we will test AI technologies in constrained environments and monitor their operation after deployment.

    I don’t even care that much if Google wants to host a chatbot, but they keep trying to imply it has safety properties it doesn’t. It’s like writing a web framework without any HTML or SQL sanitation support and saying “We will continue to develop and apply strong safety sand security practices…” and acting shocked when all the websites get hacked.



  • The lack of a speedometer in front of your face is a pretty glaring quality issue. You have to turn your head to look at the massive touchscreen. The one that replaces all other dashboards and most tactile controls for manufacturing cost savings to be a cool futuristic vehicle of the future.

    I won’t even start talking about the whole CyberStuck thing again because that’s too easy; except to point out that it has turn buttons on the steering wheel instead of a turn signal stalk, a shifter on the ceiling, and the steering wheel is not round.

    I only recently bought my first car and it’s just old enough that it didn’t even have a backup camera until I got one installed. Honestly half the reason of buying used was so I could have a car without a touchscreen haha.



  • Yeah it’s sad. As the article points out similar incidents have happened repeatedly. Anyone who saw the door design could have (and did!) predict something like this would happen. My coworker was trapped in his Tesla in his garage for 15 minutes (and he wasn’t in a panic).

    Look at the picture of the manual door release here: It’s pretty well hidden, you reach in and pull up on the door buttons.

    … then scroll down and look at the picture of the rear door manual release. You have to pull off some trim from inside the pocket, pull off another panel, and then pull a cable.

    … but wait! There’s more!

    Note: Not all Model Y vehicles are equipped with a manual release for the rear doors.

    Jesus, I hope the engineers who signed off on this think about what they’ve done and do better. I would say I hope someone regulated bad emergency door releases out of existence but… y’know.