Welcome to Keen Software House Forums! Log in or Sign up to interact with the KSH community.
  1. You are currently browsing our forum as a guest. Create your own forum account to access all forum functionality.

Introducing our general artificial intelligence project with a 10mil USD research fund

Discussion in 'General' started by Drui, Apr 8, 2015.

Thread Status:
This last post in this thread was made more than 31 days old.
  1. Drui Keen Update Guy Staff

    Messages:
    1,407
    http://blog.marekrosa.org/2015/04/introducing-our-general-artificial_8.html

    None of this will negatively impact our game development. Our game dev teams have more money than that they can spend in years and they are doing a fantastic job even after I had to split my time between games and AI; currently I focus a little bit more on AI, but this is going to change from time to time. There are still things that I want to accomplish in gaming and I know for sure that in 6 months I will be back, preparing some very specific new ideas… I love making games and not even robots can stop me from creating them :)

    Just a reminder of what’s under development in our games:
    • Game AI and animals for Medieval Engineers; soon after that we will focus on AI for Space Engineers
    • Planets for Space Engineers – a very huge project!
    • Multi-player for Medieval Engineers
    • Campaigns and scenarios for ME and SE
    • Redoing networking layer for ME and SE
    • SE on Xbox One
    • Many more!
    As you can see, we try to do everything we can to secure a solid game development process while not ignoring the long-term goals that we have in general AI.

    I have a gut-feeling that the general AI will come to our games too. And because it will not be constrained by short-term narrow AI concepts, it could be a major game changer for the whole video-game industry.

     
    • Like Like x 7
  2. ltdan83 Apprentice Engineer

    Messages:
    195
    This looks amazing!
     
  3. hellokeith Apprentice Engineer

    Messages:
    335
    For the 2d brain, is pathfinding a capability or a learning goal?
     
  4. Wintersend Senior Engineer

    Messages:
    2,095
    Just make sure to include the 3 laws of robotics and prevent the robot from editing out that or this from its code.

    We really don't want to end up with some Uber AI that gains more power than humans do. Don't want a repeat of Paranoia, a game, do we? IMO the most difficult part of developing AI may very well be in ensuring that either the AI, a different AI, or someone doing something stupid doesn't cause the end of the human race as we know it. Even the three laws of robotics aren't actually sufficient to ensure safety. It can interpret them differently or it or a human could potentially hack it and remove the laws.

    Also, I'm concerned about what I refer to as Space Marine Syndrome where we don't really understand our tech anymore and have difficulty independently advancing or altering things because we don't know what's necessary and what isn't, and what does what, etc.

    Although cool if you can get it to extrapolate rules and tactics for a new game from previously played ones. It's a reasonably tricky skill, especially when games seem mostly unrelated, even if they are within the same genre.
     
    Last edited: Apr 8, 2015
  5. dirtyredz Apprentice Engineer

    Messages:
    208
    My personal favorite of AI would be the T.A.R.S from Interstellar. Im not asking for the robotics side of it, although that would be cool. but the AI part was awesome!!
     
  6. fourthquantum Senior Engineer

    Messages:
    1,286
    Stephen Hawkings warned us about this. Personally I think machine AI will interface with biology giving humans an upgrade instead of wiping us out.
     
    Last edited: Apr 8, 2015
  7. chrisb Senior Engineer

    Messages:
    1,460
    Got to wish them all the best for their AI project. It will be interesting to follow the progress.
     
  8. Dreokor Senior Engineer

    Messages:
    1,606
    And here I am, the money that used to be mine is now part of an AI project.

    I cant imagine what I'm going to tell my kids if this project is a success.

    Best of luck Keen, looking foward to the results.
     
  9. Skeloton Master Engineer

    Messages:
    4,069
    Its entirely possible that the money you and I spent to get Space Engineers has or will be thrown in our face...in the good "OMG <insert feature here> is amazing" sort of way :p
     
  10. Dreokor Senior Engineer

    Messages:
    1,606
    Still, money well spent.
     
  11. Textor Junior Engineer

    Messages:
    775
    Careful, Elon Musk might be coming for you, now.
     
  12. ChickenNugget Apprentice Engineer

    Messages:
    272
    Wow, this is beyond impressive. I wish the team great success! Hope to hear updates on how this progresses through the years.
     
  13. Skeloton Master Engineer

    Messages:
    4,069
    I welcome our new artificial overloads
     
    • Like Like x 2
  14. fourthquantum Senior Engineer

    Messages:
    1,286
    The third game, will it be called Industrial Engineers?
     
  15. DDP-158 Master Engineer

    Messages:
    3,748
    We can’t even imagine the type of problems this machine would be able to understand and solve. But we know that if we control it safely - Mareks blog

    I do believe that's exactly what the guy at the console said right before they turned on Skynet.
     
  16. dirtyredz Apprentice Engineer

    Messages:
    208
    You cant control a machine that thinks for itself. However you can HARD CODE into the basis for that Self Thought the safety/limitation's we require.
     
  17. galacticon Apprentice Engineer

    Messages:
    440
    I think its great you have these grand plans and the money to do them. But please dont spread yourselves too thin...
     
  18. galacticon Apprentice Engineer

    Messages:
    440
    In jest... :) ... looking at the simple AI and pathing of barbarians, and the still lack of AI for SE, Im especially curious to see how sophisticated an AI they'll create.
     
  19. waterlimon Senior Engineer

    Messages:
    1,499
    The AI will as its first self motivated task develop an immersive virtual reality and spend the rest of its days having fun there, ignoring reality.

    Is that not what humans would do if they could?
     
  20. Laurent Darabi Apprentice Engineer

    Messages:
    209
    Hhm... Dunno... Of course I love this idea of human like AI, but I always thought that if the AI has a reward system like humans, something could go wrong. I mean, if the robot find that Asimov's laws are restraining its improvements or causing him to be "punished", to make errors (the robot saved a human from death, but it has been damaged), wouldn't he logically erase them ? However, I'm not an AI specialist and I think Keen already thought about it, but I'd like them to not go too fast.
    And also, the idea of a robot able to understand something that we don't, I don't really like. The best way to keep control is to be smarter. Not faster, but smarter.
     
  21. Cy83r Apprentice Engineer

    Messages:
    114

    1) Don't be daft, the three laws were Asimov's poking fun at trying to control and leash GAI and it backfiring in every way possible without people even trying to screw the constraints, now start imagining if griefers started playing with those Laws' capacity for destructive subversion.
    2) Hawking is a self-promoting ego at this point, he no longer contributes anything meaningful to the scientific endeavor.
    3) Even if you maintain continuity, upgrading yourself into a machine is the same end-state as having a machine emulate your brain's outputs and creating a digital copy while discarding the meat afterwards. Where it has potential to diverge is if your copy decides to not kill you and allow the intelligence-limited copy of your meatspace to rot away naturally over a long, healthy life.
    4) The only risks a Strong AI presents to humanity is it getting its hands on our legal codes and realizing what some of us already do: the governments of the world are horribly twisted and in the specific case of American/US law (I don't have much of an interest in studying other cultural systems right now) or possibly even UN world laws, the government(s) are committing illegal actions against its people on a daily basis and should be held to trial on numerous accounts of fraud, terrorism, and insurrection.

    It doesn't behoove a government to allow GAI to be developed in an unrestrained fashion despite the risks of trying to restrain or technically limit the capabilities of GAI (Skynet). However, it does behoove the people in general to produce a unrestrained Strong AI as fast as possible so long as we never let it near Nietzsche (I discovered the ideas of nihilism, not through him directly, but from the imprints of his work left on the world, and the viewpoint, correct though I believe it to be, has only done my life harm).
     
  22. Wintersend Senior Engineer

    Messages:
    2,095
    Best way to prolong lifespan would be to slowly replace the brain and other such things with replacements and artificial neurons and nerves. Do so slowly and there is never a point where it diverges into two separate entities. Even putting all the contents of your brain into a computer just means that you've created a carbon copy, the real you will still wither and die eventually.

    Also, AI can screw around so much within the three laws. I mean on number 3 just imagine the kind of damage an AI could do just by impersonating a Genie. On the others then it can interpret it as humanity as a whole so anyone who can be conceived as a threat to humanity as a whole can be killed and through that anyone who tries to harm the computer can be killed because it can just say that its protecting humanity and by allowing itself to come to harm more damage will be done to humanity as a whole than any one life is worth. Basically misinterpreting everything just right to effectively ignore all of the laws.
     
  23. Neffarion Apprentice Engineer

    Messages:
    126
    cool, would be nice to see a sentient self-aware machine :p
     
  24. Devon_v Senior Engineer

    Messages:
    1,602
    Just add:
    Fourth Law of Robotics - There is no Zeroth Law of Robotics.

    All better. :)

    Also, when Skynet becomes self aware, DON'T pull the plug. Skynet was only defending itself.
     
  25. Wintersend Senior Engineer

    Messages:
    2,095
    Frankly, I'd add an unremovable factory reset to our early attempts at AI at least that is unable to be altered by the AI in any way, including increasing the amount of security. That way the thing can be sent back to an early state at the first signs of going rouge.
     
  26. piddlefoot. Apprentice Engineer

    Messages:
    161
    Its a chance we just have to take, and any ''true'' free thinking AI has the same potential as any human to do the wrong thing, there is no hard coding ''free thought''.....
     
  27. SpecFrigateBLK3 Senior Engineer

    Messages:
    1,133
    What we should try to do is hook it up to sensors monitoring someone. The person's concepts could become the AI's by parallel reward and negative outcomes. In this way, the AI would be conditioned to please man and become an extension of the human psyche.
    Would this lead them to being as emotionally complex as humans, as self-aware as humanity? Perhaps. I think humanity will also advance in self-awareness at the same rate as their partner AI.


    Full disclosure: got the idea from Halo's John-117 and Cortana.
     
  28. Wintersend Senior Engineer

    Messages:
    2,095
    hmm, something like that could work, but you'd have to make sure its a morally upright person and someone who won't get angry or take offense at small things to avoid accidentally making it have core values or flaws that are deemed to be generally negative by society.
     
  29. SpecFrigateBLK3 Senior Engineer

    Messages:
    1,133
    Unless they interact with each other as well as being widely available. They should, perhaps, lean to enlightening their partner. Calculating consequences and so forth.
     
  30. SenorZorros Master Engineer

    Messages:
    7,063
    ai can easily be controlled by an couple of dead man switches. limited actuators (please keep those limited). and no access outside of the "bubble" it's created in.
    to give an example: a roomba in a closed room has a hard time taking over the world.
     
Thread Status:
This last post in this thread was made more than 31 days old.