• 0 Posts
  • 16 Comments
Joined 1 year ago
cake
Cake day: June 16th, 2023

help-circle

  • While I’m not against an anonymous stand for what’s “right”, that really was the tipping point for a lot of changes on 4chan.

    It really fuelled the idea that anonymous should have some sort of goal of justice rather than just doing things “for the lulz”. It normalized the concept of shamelessly bringing your internet culture of choice out into the real world regardless of appropriateness (most of the protests were really just 4chan irl meetups, not really protests).

    The biggest change was the sheer amount of public attention it drew to the site. That brought in a huge influx of new users who didn’t care to conform to the existing board culture (for better or worse). Things changed considerably following all that mess.




  • That’s a combination of too simple/short in your sentences, mixed with too specific jargon with no clarification. It’s dumb as hell that people don’t know stuff like what a server is, but if they don’t you have to abstract it more.

    My go to is some form of: I’m in IT, I do systems administration. I help keep all the things behind the scenes working so that everyone’s stuff works at my workplace. Less of making your email work, more of making everyone’s email work.

    Obviously I work with a hell of a lot more than just email. I’m mostly scripting out custom automation jobs to bridge gaps in the integrations between different systems. But like you said, keep it simple.


  • So for those not familar with machine learning, which was the practical business use case for “AI” before LLMs took the world by storm, that is what they are describing as reinforcement learning. Both are valid terms for it.

    It’s how you can make an AI that plays Mario Kart. You establish goals that grant points, stuff to avoid that loses points, and what actions it can take each “step”. Then you give it the first frame of a Mario Kart race, have it try literally every input it can put in that frame, then evaluate the change in points that results. You branch out from that collection of “frame 2s” and do the same thing again and again, checking more and more possible future states.

    At some point you use certain rules to eliminate certain branches on this tree of potential future states, like discarding branches where it’s driving backwards. That way you can start opptimizing towards the options at any given time that get the most points im the end. Keep the amount of options being evaluated to an amount you can push through your hardware.

    Eventually you try enough things enough times that you can pretty consistently use the data you gathered to make the best choice on any given frame.

    The jank comes from how the points are configured. Like AI for a delivery robot could prioritize jumping off balconies if it prioritizes speed over self preservation.

    Some of these pitfalls are easy to create rules around for training. Others are far more subtle and difficult to work around.

    Some people in the video game TAS community (custom building a frame by frame list of the inputs needed to beat a game as fast as possible, human limits be damned) are already using this in limited capacities to automate testing approaches to particularly challenging sections of gameplay.

    So it ends up coming down to complexity. Making an AI to play Pacman is relatively simple. There are only 4 options every step, the direction the joystick is held. So you have 4n states to keep track of, where n is the number of steps forward you want to look.

    Trying to do that with language, and arguing that you can get reliable results with any kind of consistency, is blowing smoke. They can’t even clearly state what outcomes they are optimizing for with their “reward” function. God only knows what edge cases they’ve overlooked.


    My complete out of my ass guess is that they did some analysis on response to previous gpt output, tried to distinguish between positive and negative responses (or at least distinguish against responses indicating that it was incorrect). They then used that as some sort of positive/negative points heuristic.

    People have been speculating for a while that you could do that, crank up the “randomness”, have it generate multiple responses behind the scenes and then pit those “pre-responses” against each other and use that criteria to choose the best option of the “pre-responses”. They could even A/B test the responses over multiple users, and use the user responses as further “positive/negative points” reinforcement to feed back into it in a giant loop.

    Again, completely pulled from my ass. Take with a boulder of salt.











  • Good job being so smart, mama’s little smart man! You still have to eat your veggies before you can have any dessert though!

    More seriously, the overwhelming majority of businesses use Windows as their end user facing desktop OSes. You’re legitimately just being a myopic asshat if you think that Windows can’t be trusted for anything important. (Inb4 you bring up Crowdstrike, which wasn’t a Windows specific issue, but a “we have code running at kernel level” issue, and hit Linux roughly three months prior to the big clusterfuck)

    Also, your bit about $150 cost for the OS is dumb too. The average user is buying a prebuilt with the OS preinstalled. Technically they are paying for it, but it’s a wacky discounted OEM license fee baked into the full cost. Anyone not buying a rig with Windows preinstalled can use it unlicensed, can transfer license from pretty much any older Windows OS install from the last 20 years, can just use massgrave to activate it for free, or could go buy a discounted OEM license that they can only install to one machine. The full price license allows for install on multiple machines, which you don’t really need.

    My point is, very few people are paying full price for a Windows license.

    Full disclosure, I agree that Microsoft is a shit company. But this elitist shit is just stupid. Especially when it’s almost pure posturing.