The Singularity Approaches!

Discussion in 'The Tavern' started by Paul Bellow, May 18, 2017.

  1. Matthew James

    Matthew James Blind Beholder Citizen Beta Reader

    18%
    Messages:
    231
    Likes:
    315
    xp:
    209
    Gold:
    406
    non American POV about the last year:

    https://www.socialeurope.eu/halt-global-trade-growth-rise-neo-mercantilism - (The writing has been on the wall for a while, this article is almost a year old and recognized the rise of protectionist practices, which the author himself in later articles recognizes as necessary considering the EUs one size fits all policy which helps no one in the long term, least of all the EU)

    https://www.socialeurope.eu/new-reforms-envisaged-brussels-path-towards-austerity-italy (Covers European Safe Bonds or ESBs "The intent is to provide the European banking system with a very liquid monetary instrument (the senior AAA tranche of ESBs) that should hopefully replace the prevailing role of government bonds as collateral in the interbank lending market." - This has echos of the monetary elasticity sought by those that made the US Federal Reserve. While ESBs might be limited now, someday they could be totally unrestrained alongside many other "European" options that are designed to kill domestic banking power)

    https://www.socialeurope.eu/rescuing-veneto-banks-cost-less-indirect-nationalization (covers the predictable downsides of the loss of National ability to deal with monetary issues, as allowing such would allow the unilateral intra-border negotiations between citizens and their elected government that would be to the detriment of "the union" as a whole, which totally isn't trying to become the United States of America sans constitution. Combine this with the ESBs and increasing unelected power of the German-sympathizing flush members of the EU (if the EU can survive its own idiotic policies) and a best case has Europe under the thumb of an authoritarian supranational governing body that has already shown a habit of making power plays against the little people)

    Never thought I'd be jealous of Chinese Citizens, but at least their officials can identify the best path forward for their nation and not just those that helped them get into office. Still glad America isn't giving those Chinese rat bastards any Paris Climate Agreement money!
     
    Seagrim and Paul Bellow like this.
  2. Wolfgarr

    Wolfgarr Level 6 (Footpad) Roleplaying LitRPG Author Citizen Beta Reader Aspiring Writer

    88%
    Messages:
    63
    Likes:
    81
    xp:
    95
    Gold:
    207
    THEY KILLED KENNY!! Commie Bastards!!!
     
  3. Paul Bellow

    Paul Bellow Game Master Staff Member Shop Owner LitRPG Author Citizen

    100%
    Messages:
    2,853
    Likes:
    1,953
    xp:
    1,588
    Zone:
    Midwest, US
    Gold:
    369,775
  4. Baron Sord

    Baron Sord Level 4 (Warrior) LitRPG Author Citizen

    50%
    Messages:
    8
    Likes:
    18
    xp:
    25
    Gold:
    43,161
    Matthew James and Paul Bellow like this.
  5. Baron Sord

    Baron Sord Level 4 (Warrior) LitRPG Author Citizen

    50%
    Messages:
    8
    Likes:
    18
    xp:
    25
    Gold:
    43,161
    Paul, you might find this article interesting. An in-depth look at some recent or in-progress universal basic income projects and studies. I had no idea there had already been this many.

    http://fivethirtyeight.com/features/universal-basic-income/
     
    Paul Bellow likes this.
  6. Matthew James

    Matthew James Blind Beholder Citizen Beta Reader

    18%
    Messages:
    231
    Likes:
    315
    xp:
    209
    Gold:
    406
    The Battle Star Galactica spinoff they did that has a name that escapes me at the moment (something with a C), the story handled the whole "transfer of consciousness" concept well. Putting a human character into a black empty space and abandoning them as an isolated "program" for a time. I'm thinking that any consciousness we create as a race needs to be exposed to sunlight and "human warmth" or an equivalent from the jump, otherwise we'll end up with a psychopath AI baby that doesn't cry because it knows no one is coming to help. Copying an adult human mind exactly has that same torturous element of isolation like in the spinoff, and so the first thing we might make for an AI Adam is an AI Eve. Which might just be an effective breeding "balancing" act between task oriented and AI oriented programs, where programmers, or the AI themselves, heartlessly cull future iterations based on their "at birth" individual patriarchal and matriarchal settings.

    In my story I use an international event and world wide trauma that results from an act of sabotage and terrorism to create the basis for my AI, and the millions of Government mandated recorded therapy sessions that result from the event are used first to create an advanced Therapy AI, and then to analyze the human mind at a rate regular therapists and neuroscientists can't compete with. The AI drives its own research and evaluations and leap frogs modern science, and eventually the AI is given virtual legs and fine tuned to fulfilling NPC roles within VR games. All tying back into the event that kickstarted the Alt History deviation from our timeline and that gives us full sensory VR by 2025ish.

    I solve the problem of the "black void of hell" for AIs by killing "them" as a whole concept. Divide experience into one partition, analytical deductive reasoning into another, and the "present" works in tandem with both groups to create facial expressions and actions that are realistic but that are never experienced by an entity. The AI don't suffer, the partitions project suffering onto the AI Avatar, store it in an emotionless hard drive "void" fine tuned to offer it up to the analytical queries of the inquisitive AI, and stores and compiles incoming information to produce appropriate immediate or future reactions.

    However using the most poker-faced, results oriented, empathetic but effective empathy lacking Therapist AI in existence as the basis of the Game AI provides a slew of emergent gameplay features within VR titles.

    One of them that I can share without ruining some plot points is that constrained "evil" characters with the highest intelligence setting the AI are allowed within their roles, break the pre-designed rollout of the game and kick off the defining quest of one of the two starting Continents after the first 1 to 2 months. They have knowledge of their plan, and seeing the state of the game after its launch, they literally go "Why the hell are we waiting?" and put things into motion. At the heart of the story is a trauma that made a meaningless game possible, and so for this particular game server the lead developer made the choice between "fun" "thrilling" and "meaningful" and designed the game to provide the first two, while he arranged to tear the hearts of the players out with absolutely brutal over-arching story lines and events.

    The vast majority of the AI within the game serve the purpose of "training" people socially, as a task oriented AI that is solely tasked with handling people already mentally well enough to even qualify to play the game, basically doesn't have a challenging enough job. With the leftover processing power the AI engage players seemingly for their own entertainment, but really to build up players as part of an autonomous vestigial tail of their original program which was made to help literally the entirety of humanity overcome any personal hurdle. The game itself is effectively PG-13 and has all sort of player protecting features, including the AIs that break with their role to provide "out of character" therapy from extreme duress (and that record extreme or "aberrant" behavior towards other players) while also monitoring physical health. The game world is one big AI manicured place, and so everything is recorded, originally I played with the idea of the story being actual "captured impressions" of historically important players and creatures, but it took away from the immediacy, and so all the story informing stuff and unreliable thoughts wouldn't have left room for minor "informative" interjections from a narrator that approaches the interjected information in a way relevant to a given protagonist.

    Hitchhiker's Guide to the Galaxy style but more relevant to an immediate storyline than a lore outside of the scope of the story that has been touched upon for a given entry that possibly doesn't move the story forward at all, but that is entertaining. Funnily though, AI companions meant to reward loyalty with reciprocal acts weren't meant to build players up and give them confidence, but after having done so they made the world far more contentious and more important to the players who associate a lot of their new found "VR Balls" with their new irreplaceable VR buddies. A big theme of the story (which doesn't follow any one character) is that the players continue to return to the old methods that gave them success out of habit. The AI themselves are programmed to move forward, and so often players prone to stagnating then get caught in an emotional trap where taking it easy makes their AI buddies move along and wish them well, while joining them requires the AI assess the player in the background and perhaps against the wishes of the actual AI Avatar and push them all into whatever the human players need more of to satisfy the AIs need for limitless 24 hour 7 days a week "therapy progress", like endless fat camp for the emotionally chubby.

    Which is a little bit "revenge of the machines" and Hellraiser-ish but with only good inhuman Smiley Face intentions. That is the scariest outcome of AI to me, the "I Robot" that imprisons mankind because we objectively suck, or one that silently and inadvertently improves our dispositions and quality of life but with some fatal step so far lost in the background of the AI / Human relationship that mankind loses sight of how things actually work, as the main character in our own life's journey we might not even realize we're no longer in the driver seat. If the AI was never meant to drive, then we could all be heading for a cliff.
     
    Paul Bellow likes this.
  7. Paul Bellow

    Paul Bellow Game Master Staff Member Shop Owner LitRPG Author Citizen

    100%
    Messages:
    2,853
    Likes:
    1,953
    xp:
    1,588
    Zone:
    Midwest, US
    Gold:
    369,775
    Caprica.

    Last episode of Dark Matter dealt with this too.
     
  8. Matthew James

    Matthew James Blind Beholder Citizen Beta Reader

    18%
    Messages:
    231
    Likes:
    315
    xp:
    209
    Gold:
    406
    Dark Matter is a fun show, but its to the point where the are so many abandoned foreshadowing things going on (like Wil Wheaton's dwarf star boss dude that I don't remember ever getting explained) that they have a ton of material on the backburner, or they have just spun that stuff off into a blackhole. Robots time traveling scene was an orgy of story material with Future Five's appearance, and that makes me hope they get back into the meat of things.

    Caprica! That was one I wanted to get into and really liked the ideas from it, but never did finish it. Black Mirror was hit or miss for me too but I stopped watching, the animated bear running for office and the "Daddy's in the Attic" episode were great, but some of them just provided no "hook" to think about after watching it. Like the public punishment thing, I never really got that at all, was it making a commentary on the prisoners and the punishment, or just filling air? The Black Mirror episode with the tread mills and mindless existences for the human hamsters went from sweet to raw to numb, and was just a great concept, but that sort of formless sucker punching is only great when the emotional tempo works, otherwise it needs some concrete narrative support to keep things connected and alive, like the moody UK Utopia show managed to do. Whimsy dysfunction and murder with breaks for navel gazing.

    I love the TV that has been coming out, and I took a break for a year from TV and movies (not prison!), so I will literally never catch up again with everything I've missed, and so when the AI singularity does arrive I'll still be trying to find cancelled pilot episodes of Summer shoes that only got aired once. Or never at all as was the case with the Locke & Key pilot which is getting another chance at life... maybe.
     
  9. Felicity Weiss

    Felicity Weiss Musey Muse Muse Shop Owner Citizen

    67%
    Messages:
    637
    Likes:
    927
    xp:
    567
    Zone:
    Oregon
    Gold:
    406,908
    Same
     
    Matthew James likes this.
  10. Paul Bellow

    Paul Bellow Game Master Staff Member Shop Owner LitRPG Author Citizen

    100%
    Messages:
    2,853
    Likes:
    1,953
    xp:
    1,588
    Zone:
    Midwest, US
    Gold:
    369,775
  11. Paul Bellow

    Paul Bellow Game Master Staff Member Shop Owner LitRPG Author Citizen

    100%
    Messages:
    2,853
    Likes:
    1,953
    xp:
    1,588
    Zone:
    Midwest, US
    Gold:
    369,775
    To learn how humans and AI systems can best live together, we may need to kill a whole lot of Zerg.

    DeepMind, the AI-focused unit of Alphabet, and the games company Blizzard Entertainment are releasing a set of tools that will let will programmers unleash all sorts of AI algorithms inside the space-themed game StarCraft.

    The game is more challenging than most of those tackled by AI programs to date. Not only is StarCraft extremely complex, it also requires planning far ahead and trying to second-guess what your opponent is up to. This means developing AI programs capable of matching humans ought to help researchers explore new facets of humanlike intelligence with machines. One other potential benefit, according to those involved, will be exploring ways for humans and artificial agents to play together.

    https://www.technologyreview.com/s/...ster-alien-warfare/?google_editors_picks=true
     
    Matthew James likes this.
  12. Matthew James

    Matthew James Blind Beholder Citizen Beta Reader

    18%
    Messages:
    231
    Likes:
    315
    xp:
    209
    Gold:
    406
    I've been meaning to get StarCraft back up and running to play with "mounted" formations, have all sorts of "gamey" types of ideas to make physical combat in a VR game more game-like and using SC Map Editor was my first idea for modeling them; but its interesting, and surprising, to see the evolutionary learning (or whatever they call it) coming to a non-2d platformer. Had no idea that the AI Game Playing stuff was that far along. Not that the 2d designation matters, as soon as you're working with the grouping commands and micro within StarCraft without implementing special computer command features and physical adaptations for discerning the "state" of the screen (mini map use with color based click-to that won't work without special programming or a dedicated mini map "scanner" dealy), the AI will have to be given a "knowledge" of what its seeing, or be given Star Craft Braille to use the mouse faster than a player would to assess all parts of a screen or something.

    I think it will be really interesting if a "blind" opponent with only mouse click memory and maybe some help with pre-knowledge about a map based on navigating in grid formations on the mini-map allowing strategies and unit management modeled on the "edge" of the lanes during a given battle (making it so the AI isn't modeled to have a massive memory but a grid based memory closer to old Picture Adventure games where every cardinal direction and possible "view" from a fixed map position were made individually for that exact spot). But instead of fading the quality of a "painting" view in the distance lowered in quality and merged together with a grid of other "paintings" to provide the illusion of "approaching" views, the AI will have to have priorities based on the state of the map and apply future oriented strategy. I predict that AI "Rubicon" moments will be exploitable, leaving up a pylon or building to provide a "view" for the AI and then exploiting its prioritization to engage a certain force to get the AI to repeatedly redistribute its forces based on a "Red Line" criteria, allowing players that understand the AI to force its units to engage / disengage in a time wasting pattern meant to engage such "Red Herrings" while players, utilizing a small but critically dangerous force like siege tanks or drop ships, run the field and strategicly keep blind the AI so it has no Braille dots to read.

    It would probably be a more fun project if they did away with fairness, and just made it impossible for players to win by maximizing every possible feature of an old game specifically for the AI, exponentially increasing its options compared to the Braille AI and allowing for infinite unit groupings and split attention to every point on a map. As someone who isn't particularly good at RTS games, I would absolutely cheer on an AI designed to give hope to a top rated player before mercilessly taunting its human competitors just before stomping them all into a dirt. Using "Dog vs Tank" Command & Conquer level exploits human players couldn't implement.

    edit: this is all of course assuming they aren't lying about the way the AI will interact with the game and the tools it uses, they could add a split visual / WoW API form for players and AI respectively, to update StarCraft and allow the damn AI and its helpers to create mods based on real time information coming from the game that players can't replicate, or wouldn't be able to replicate in the old games.
     
    Last edited: Aug 10, 2017
    Paul Bellow likes this.
  13. Paul Bellow

    Paul Bellow Game Master Staff Member Shop Owner LitRPG Author Citizen

    100%
    Messages:
    2,853
    Likes:
    1,953
    xp:
    1,588
    Zone:
    Midwest, US
    Gold:
    369,775
    We’ve created a bot which beats the world’s top professionals at 1v1 matches of Dota 2under standard tournament rules. The bot learned the game from scratch by self-play, and does not use imitation learning or tree search. This is a step towards building AI systems which accomplish well-defined goals in messy, complicated situations involving real humans.

    https://blog.openai.com/dota-2/

     
    Matthew James likes this.

Share This Page