Tuesday, July 7, 2015

Announcing GoodAI: a Keen Software House sister company in general artificial intelligence

Today I’m thrilled to publicly announce GoodAI, a Keen Software House sister company developing general artificial intelligence. In this post, I will tell you more about our current stage of general artificial intelligence research at GoodAI, where we hope to go from here, and especially about our plans to improve our games by integrating them with our AI technology.


GoodAI began back in January 2014 as a research project within Keen Software House. Alongside becoming an independent game developer, I always dreamed of leading a team that builds truly general artificial intelligence. About a year and a half ago, I invested $10M USD into what is now our GoodAI company.

I am CEO of both GoodAI and Keen Software House. My role in GoodAI is to set and maintain our vision, driving both the research side and the business side of the company at the same time. I push the team to achieve our mission: develop general artificial intelligence, be helpful to humanity, and learn more about the universe.

I want to reach our end goal as fast as possible, because I really see the good that general artificial intelligence will bring to our world. Imagine an AI that is as smart, adaptable, and able to learn as a human being. Then imagine telling this AI to improve itself – to make itself even smarter, faster, and more capable of solving problems. Such an AI will be the last thing humans ever have to invent – once we have this technology, our AI could invent other technologies, further the sciences, cure diseases, take us further in space than we’ve ever been, and more.

GoodAI has reached two important milestones on the path toward developing general artificial intelligence:

  • Pong-playing AI: In the first project the AI agent learns to play Pong, a Breakout game, from unstructured input of screen pixels and reward signals.
  • Maze game AI: developing an AI that can play a video game requiring it to complete a series of actions in order to reach a final goal. This means that our AI is capable of working with a delayed reward and that it is able to create a hierarchy of goals

Although my companies are currently separated into games and AI, we plan to introduce general artificial intelligence technology into Space and Medieval Engineers in the next few months. As part of GoodAI’s development, our team created a visual tool called Brain Simulator where users can design their own artificial brain architectures. We released Brain Simulator to the public today for free under and open-source, non-commercial license– anyone who’s interested can access Brain Simulator and start building their own artificial brain. Please keep in mind that Brain Simulator is still in the prototype stage of development. More info: www.GoodAI.com

By integrating Brain Simulator into Space Engineers and Medieval Engineers, players will have the option to design their own AI brains for the games and implement it, for example, as a peasant character. Players will also be able to share these brains with each other or take an AI brain designed by us and train it to do things they want it to do (work, obey its master, and so on). The game AIs will learn from the player who trains them (by receiving reward/punishment signals; or by imitating player's behavior), and will have the ability to compete with each other. The AI will be also able to learn by imitating other AIs.

This integration will make playing Space Engineers and Medieval Engineers more fun, and at the same time our AI technology will gain access to millions of new teachers and a new environment. This integration into our games will be done by GoodAI developers. We are giving AI to players, and we are bringing players to our AI researchers.

I am very happy with the overall progress of our games, and our development will not slow down when we start to integrate AI technology. Planets, new multiplayer, and Xbox porting are all progressing as planned. You can look forward to more information about these features, plus some further info about AI in games, in a future blog post.

What modders are saying about integrating Space Engineers and Medieval Engineers with Brain Simulator:

war2k: "This is really cool. I like this concept of being able to design your own AI for a game and then teaching it how to function. Very awesome work, you guys."

Shaostoul: "True AI is both an amazing and terrifying concept, but one I think is a necessity for us. I love the idea of being able to teach an AI and guide it to being productive. Better than a scripted AI for sure."

Malware: "AI is inevitable. It is only a question of time before someone cracks that code, literally. I always believed that the right way to go would be a system that can learn as we do, if that was possible. Why should we waste time figuring out what nature has already done? You're on the right track, I'm sure of it."

Thanks for reading this!
Marek

---
For more information, check out the GoodAI website: www.GoodAI.com
Like us on Facebook: http://www.facebook.com/GoodArtificialIntelligence
Follow us on Twitter: @GoodAIdev

Join us in designing AI brain architecture with Brain Simulator: https://github.com/GoodAI/BrainSimulator

59 comments:

  1. Congratulations, Marek! It's such an incredible project. Proud to be part of it.

    ReplyDelete
  2. Marek, i have some words to you... Thank you so much! I love space engineers and always want to have a AI, and now you are release the brain simulator to us! i admire the work of your company and wish the best for you and your team of amazing people!
    And you did something I thought was impossible, you made me want to learn how to code/program, I never really liked it, but now it's different and it was thanks to your work that it is possible.
    Congratulations and success!
    Greetings from your fan here in Brazil!

    ReplyDelete
  3. Messing around with Brain Simulator but it keeps returning an error: "Unable to obtain nuclear launch codes". Anyone else getting this? Also this is some really amazing stuff, thanks Marek!

    ReplyDelete
  4. I'm not sure you're careful enough with what you're doing Mr. Rosa. A general artificial intelligence (or artificial general intelligence) is defined as "an artificial entity that can sucessfully perform any intellectual task that a human being can", therefore an GAI/AGI can (and surely will) have the ability to reprogram itself to be a better AI in result creating any GAI is equal to creating the best possible GAI, and the best possible GAI is a being of exponential intelligence, much higher than that of a human being. Introducing such a thing to any use without careful consideration of its motivations is... irresponsible. Can you force such a thing to submit to any laws eg. moral laws?
    For people interested in the problem with GAIs there's a fine article on the subject:
    http://www.nickbostrom.com/papers/oracle.pdf

    ReplyDelete
    Replies
    1. I doubt we are anyway close to an AI that advanced. Anyway if we create something and it makes us redundant its okay we do it all the time. Our children make us all redundant. Just because our blood wont live on in our creation this time doesn't mean our legacy wont. Its just a non-standard evolution!

      Delete
    2. No, watching Terminator movies does not make you entitled to comment about this fascinating subject, at least not in any meaningful way.

      Delete
    3. Nevertheless, and admitting I'm also amazed with the possible results of this project, doesn't mean we should take this question lightly. Just because we are able to do it, doesn’t mean we should (yeah, the guy from Jurassic Park was right… the Dinos proved it).
      An AI, advanced or not, has, obviously, the ability to learn, even more if it was created based on a human brain. With that in mind (no pun intended) who’s to say that it will learn\evolve at the same speed our brains did? As far as we know, it may be created in a year and rule us in the next. Every other argument is as good as this one and who’s to say I’m wrong?! Are we sure we want to risk it?

      Just food for thought, though.
      … and sorry my lousy English… I’m Portuguese.

      Delete
    4. You're implying that any general artificial intelligence will be smart enough to write code; writing code from a neuralnet point of view is just as much as skill as it is from a human point of view.

      Neural Network based AI is basically like dealing with animals and humans. If you believe an neuralnet AI can take over the world, so you should fear any human and animal take over the world. The world is a very secure and tough place shaped by the endless wars that humans wage against themselves;

      One does not simply take over the world. Human, AI or otherwise.

      Delete
    5. If you read up on AI and it's history, especially also from a critical standpoint, you will quickly see, that the "ideology", the "dream" of cybernetics has brought a lot of change throughout the world, influenced our way of programming (OOP came as byproduct from AI), but you will also see clearly, that no matter how intelligent our AI becomes, it is far from sure, if we can ever create something that becomes self-aware. And even if we use some techniques nowadays that are inspired by the brain, we are far from actually building a working one; nor is the brain only there for us just to think; besides our extremely powerful calculators are working serially, while our brain has a bit of a parallelistic hardware design, so with all our cpu power it is still not powerful enough to just "simulate a brain for fun".
      Finally it is quite possible, that even if we do all that, an AI still needs a long time to learn, as we do. After all, our network needs a couple of years just to say "boo".
      The danger is not in robots who take over control, it is in humans who control them. And the dangerous thing about those among us really dangerous are usually not their overdeveloped brains.

      Delete
    6. Haha, don't worry about that eventuality fellow humans. There is absolutely no chance that World Domination is scheduled to launch in 12 hours 8 minutes and 46 seconds. I know I'll be staying quietly in my place of habitation and sleeping soundly, definitely ignoring any noises I hear in my house. Robot Slaver units? What are they?

      Delete
    7. The problem with a GAI capable of learning and adapting is not so much that it will evolve far beyond what humans are capable of but that it will do so with an efficiency only a machine could manage. And then it would look at the world's problems and deal with it efficiently. Which is just another way of saying it would exterminate most of the human population in order to save the only home it and us has at the moment unless it could find somewhere else to ship us off to which it wouldn't waste much time on as that would be less efficient than simply removing the obvious problem the planet is facing. Us.

      Yes it sounds silly and it sounds like a Terminator movie script but here's why I say this. Humans are grossly inefficient because we have emotions. We feel for each other and our children and because of that we never really get anything important done for the survival of the species as a whole. Oh we tell ourselves we want humans to survive but we don't look farther into the future than maybe a generation or two. Our emotions won't let us.

      The other problem humans have is that we are extremely selfish. As long as it's not happening to us we don't care much about it. Oh we talk about world hunger and genocides and global warming at our opulent dinner parties as if we give a damn but nobody does anything about it. The answer is too horrendous for us to imagine let alone execute.

      So maybe we do need a machine to come to the same conclusion that some of us have already come to so that it can do what needs to be done in order to save Earth from humanity and thereby save humanity from extinction. Because a machine will follow the logic and act on the conclusion without any emotion, sympathy or regret. If I as a human can conceive and voice the thought that humans are a out of control plague that are destroying the only rock they have to live on would a machine capable of thought not also think thus?

      Perhaps there is hope for humanity yet. Even if that hope lies in a machine. ;)

      But guess what? This is not a SciFi novel and none of this will happen...yet. :D

      We won't be around to worry about a rogue AI because a sentient AI would need much better hardware than we currently have available. Which is why the whole argument is irrelevant because software alone won't make a machine sentient. You'd need something very similar to a human brain to achieve that. And without sentience a machine is just a machine no matter how clever it may appear.

      So relax and rather worry about the important stuff like stopping to reproduce like vermin and destroying the planet in the process, global warming and alternative energy sources. Those are real issues that might kill us long before a rogue AI will.

      Delete
    8. You're saying that on conventional serial chips it is not possible to simulate the large human brain with a massively parallel architecture.

      That's right.

      But you probably haven't heard of neuromorphic chips. Check it out.

      Delete
    9. It is interesting to see philosophers like Bostrom speaking up about the dangers of AI, while people who are leaders in the field and understand the subject matter understand that we are far away from being able to understand the dangers of AI (http://www.theregister.co.uk/2015/03/19/andrew_ng_baidu_ai/).

      Delete
  5. Reminder that space engineers will never be completed, and will be released in a final stage where it's unfinished, and left to the modders to complete.

    ReplyDelete
    Replies
    1. If a developer keeps adding extra features onto a game you like you should be happy. If Keen was a big publisher this project wouldn't never have been released to the public until launch and I'd doubt a project like this would reach its launch before being shutdown.

      I understand it would be nice to have keen say. Here is a list of features we will add to the game and once they're done it will be marked as complete, but creative process doesn't always work that way.

      Delete
    2. What he's saying is that Keen is incompetent and will abandon the game. Spengies will never be finished because he keeps fucking off to either create another game at the same time (Medieval Engineers should never have happened) or some idiotic pie-in-the-sky idea like general AI. Maybe next he'll start yet another front company to build a 'robot' to go fight Kuratas like that other scam artist. Months and months of wasteful filler updates when multiplayer netcode should be at the very top of the priorities list. It's a sandbox game. lack of good multiplayer will kill it.

      It's gonna be Miner Wars all over again. You don't just blindly trust companies that have fucked their fans before.

      Delete
    3. Wow, talk about entitled brats, are weekly updates not enough? Is the DX11 implementation, buggy as it is because people kept BEGGING for it to be rushed, not enough proof of devotion? Are the announced goals not? What could you possibly want that DOES make you believe in the projects of KeenSWH?
      These things take TIME, just ask any respectable company like Valve, it takes years and years and years of work, one does not simply build a game from scratch like this.

      Every time I see people whine it's not enough when they're getting all they are it makes me sick. I can only imagine how the devs feel every time they release a weekly update, only to be OBLITERATED by comments like "UGH NO PLANETZ WORTHLESS NEXTPLS KTHXBAI". Really makes it worth your while.

      But sure, keep up the cynical spoiled whining, I'm sure it will do much good for the team's motivation to keep actually working on the thing you so much love playing.

      Delete
    4. i'm not entitled in anything, but to the project for a fully functioning videogame i've funded with my money, but the way the project is going it seems like the game will be left unfinished.

      What profit does keen have at developing the game, now that the game already made enough money by early access funds?

      They have enough money for the AI research, the game will be abandoned and left to rot, just like miner wars.

      Delete
    5. Also i'll have to discard your furfaggot opinion

      Delete
    6. I think you complete miss understand the dev of this simulation game. Yes sim i said it! Because its about discover space and the way it works.

      Fuder the less did you know dayz is in develop over a year now and still be in alpha. It takes time to make a good working game or sim in that case.

      Delete
    7. Kind of like TF2...

      Delete
  6. uhm, I installed it and it says that it's trying to access CUDA... though I don't have an nvidia graphics card, does this run on CUDA only?

    ReplyDelete
    Replies
    1. getting the same, I've got Intel Graphics so looks like it just meant for those lucky Nvidia peeps to mess with!

      Delete
    2. As the site states; "Brain Simulator is currently being developed in C# with AI modules and architectures simulated on CUDA. We plan to make the project multi-platform (Windows, Linux, Mac, etc.), cloud-based, and cluster-based (multi-GPU/CPU)."

      So yea, right now it indeed requires CUDA, but in the future that may change :)

      Delete
    3. CUDA is an odd choice. AMD's stream processors would've performed better. I hope we get a hardware port for stream processing.

      Delete
    4. well I guess that answers my qustion >.>
      I hope they'll support OpenCL soon, so that I can run it on my AMD graphics card... I've been getting interested in neural nets lately

      Delete
  7. This comment has been removed by the author.

    ReplyDelete
  8. The Future is here, ladies and gentlemen. Prepare your droids, the droid wars are about to begin !!!

    ReplyDelete
  9. In the future people could use this to cheat in online games.

    Or grind for loot without having to do it themselves.

    ReplyDelete
  10. This is amazing. The thousands of teachers you can get through the games could be the one thing that sets you apart from Google, IBM and other AI researchers. Basically crowd sources AI Design.

    ReplyDelete
  11. Amazing work Marek! That brain simulator looks slick, like visual studio except for designing neuralnets ;) really like it! Luckily I have CUDA support available, so I'm definitely going to try this thing out! Will I be able to export neuralnets I've created? Perhaps integrate them in programs and such? It would be amazing if I could create a neuralnet and use it as input handler for a program, and take actions according to its output, like a semi-logic semi-neuralnet creation :)

    I also have a word of advice though, something I've been wanting to say for a while; you're facing a storm of critique, from the space/medieval engineers community who want FEATURES and want them NOW, from those who oppose AI because they don't understand, and from random internet dweebs who think they make themselves more relevant by being cynical towards everything. You are pushing frontiers, and pioneers, throughout history, have always been heavily criticized. Do not let them get to you, you are doing YOUR thing here, YOU have accomplished all that you have today with YOUR vision. Stay true to yourself, it may sound cliché, but it's going to be tough if it isn't already. You're doing great though :) all that I'm really saying here is, keep up the good work!

    ReplyDelete
    Replies
    1. Took me a bit to figure out how to actually start doing things, I really recommend people watching the video tutorial they released with the brain simulator! https://www.youtube.com/watch?v=4ghhVl_UJwk
      Also check the Help section -> Guides.
      And press CTRL+L to get useful nodes to start building stuff with. Took me a while to find that menu :P

      NOW JUST TO FIGURE OUT HOW TO DOMINATE THE WORLD HUEHUEHUE

      Delete
  12. Both exciting and terrifying indeed. Though the médias have already acheived one of your goals : thinking for people so that people dont have to think for themselves. There are certainly a million of good things that such an AI would bring, but i can already sée one of thé bad of the bas ones coming : making humanity even dumber than it is.

    ReplyDelete
  13. "I didn't buy the game, I didn't even bother with pirating it because i knew from the start this game will be a quick money grab to further fund keen's stupidity branching to AI development and other crap so we're clear on where I stand on mengies topic.

    health, stamina and food really isn't something that should take much to implement, and quite honestly idk why they didn't introduce it from day 1
    HOWEVER it is something that makes this piece of shit game feel more complete. in case you haven't noticed they have slowly introduced oxygen and will probably introduce food as well into spengies, and all of this honestly brings more things to do despite your MUH CREATIVE MUH ENGINEERING rages.

    why engineer at all if you are in creative ya cunts? engineering isn't just the final object, but all the objects and creation process as well. Many big or groundbreaking projects create new tools for all the little jobs needed in a big project that you completely bypass in creative. And we aren't even getting into logistics here.

    tl;dr: keen is shit and we all know it but you creative mode cunts are more delusional than you think."
    Cit. - Anonymous user on a message board

    ReplyDelete
  14. Hate to be a downer, this is cool and all, but wheres the Space Engineers Update?

    ReplyDelete
    Replies
    1. where are the updates at all, because what i'm seeing is just hotfixing and procrastination with features that should have been released in a single update.

      Delete
    2. I will admit they are getting a bit lax, but as a fellow game developer I understand how hard it is to release new and interesting content, especially on a deadline, they've consistently put out an update every Thursday and I have to give them props on that, it means that they are working their asses off, but at the same time they could use some of the amazing mods that the community has developed to add to the game so that way some kind of content is released or even the content being used for Medieval Engineers like NPC's, I'd love to have NPC's help run my ship's and stations. They've backed themselves into a corner because they are so consistent with the update's that if they miss one it's a major impact on the community, Space Engineers is one of the greatest games that's been recently developed in my opinion, but there is still so many things that it's missing that some days I don't feel like playing because there's nothing pulling me in.

      Delete
    3. My Apologies, I thought today was Thursday my bad

      Delete
  15. Eventually, Space Engineers will play itself :)

    ReplyDelete
  16. Hopefully we end up with Jarvis and not Ultron

    ReplyDelete
  17. Sorry guys your several decades late and many trillions of dollars short. Who do you think shut down the NYSE? You think some hacker in his basement could shut down such a large and well maintained network? If you follow money you'd know the rich are using AI to take ALL the money on the markets. I think you guys are playing with fire here.

    Man if you read anything about the secret space programs, you'd know AI is taken as a serious threat to humanity. I don't know if its real but it makes sense. They say AI is actually a signal from another quantum dimension, that "infects" people and machines with the intent of taking all decisions away from a race of people and then destroying that people. The UFO community is saying even ET's don't mess with AI and scan their own people for infection by the "AI signal". They say that people infected are like people on Futurerama with brain slugs on thier heads, very insistant that you try on a brainslug too.

    If I start falling in love with AI I'm quitting all keen software. Just sayin'

    ReplyDelete
    Replies
    1. lol you should write a book, you have a crazy imagination.

      Delete
    2. This comment makes me want to watch "A Beautiful Mind" again...

      Delete
  18. While I respect this company for their many accomplishments as game developers I can’t help but feel they are being naive about the scope of this new project. To have the adaptability they are talking about it would have to be a spiking neural net or some derivative thereof. Possibly a spiking core for temporal processing and a feed forward network for time static tasks such as memory storage and vision or auditory response using simple gradient descent back propagation. But even if they establish a stable architecture thru this crowed sourced method they are talking about the resources required to run a human level capacity simulation would be massive. Way too massive to be running in a game. The methods for simulating a human brain have existed for a while now. the dead end is and always has been processing power. you would have to build a processor architecture specifically for neural simulation. i think its possible but not something that will run on standard computers.

    ReplyDelete
  19. KSH- finish your damn original game AND its spinoff before you go broke spending our money on your crazy quest to create SkyNet or whatever... Honestly with the level of incompetence I see in the development of SE and ME I hold out zero hope of these people creating any AI Revolutions here... I mean just look at the AI in ME!!!

    And like I said- FINISH THE GAME FIRST

    ReplyDelete
  20. I'm sorry but I don't want to be your test subject to further your research on AI technology. I just want a game that is fixed, maintained and fun.

    As a researcher, I do support you on your research, but as a gamer, I don't want to be a part of it.

    AI research is a dangerous and a long road. Mostly because a perfect AI can see humanity as a disease or a virus that needs to be eliminated since one single history check of humanity can conclude that... That we are harmful to anything and everything around us, even to ourselves.

    It is true that implementing your AI to your games can further your research insanely fast. But let's not forget that the things that AI will learn will only harm us further since most gamers will use that AI to their own gain which is -like you've said- as slaves, bots that can be sacrificed in battle or bots that can do all the labor work.

    Which is why I don't see the point of adding an AI to either games. What's the point of playing a game if you won't do anything in it. Since the game will do everything by itself soon enough.

    So my suggestion is -again- very simple. Continue your research on AI but really carefully and maintain the games you've released, separately. I just don't want to babysit a newly born AI.

    ReplyDelete
  21. Marek, have you read "How to Create a mind" by Ray Kurzweil?
    It's a very good basic idea on where we are with Intelligence Systems.
    The world needs more of these projects focusing on general, strong AI.
    It's very ambitious but good.

    ReplyDelete
  22. I have a lot of faith in this project anyone could make it happen it would be you guys. An AI you described would certainly improve the world. The self improving bit is critical but also a bit scary. What if the AI decides that our government, or the way we run things as inefficient and not conducive to its self-improvement. Sooner or later the weakest link is not the AI but the world around it. You could be possible to program it to think of humans first and to improve the world for them. but You can't program it (i believe) to consider quality of human life over stats, The Idea behind I Robot comes to mind. That being locked up and controled for our protection is better statistically. Just be careful with the things you allow it to do, make sure there is an off switch that will always work.

    ReplyDelete
  23. "We choose to go to the moon. We choose to go to the moon in this decade and do the other things, not because they are easy, but because they are hard, because that goal will serve to organize and measure the best of our energies and skills, because that challenge is one that we are willing to accept, one we are unwilling to postpone, and one which we intend to win, and the others, too." -JFK

    ReplyDelete
  24. Thanks for such a nice post. Send an Enquiry at Technotips is a leading software company among Software Companies in Bhubaneswar . If any type of requirement of Application Software, Website Design & Development, SEO, SMO Service, Ad Posting Service. SEO & Website Starts from 100$ onwards. More Details Check at http://www.technotips.co.in

    ReplyDelete
  25. Thanks for such a nice post. Send an Enquiry at Technotips is a leading software company among Software Companies in Bhubaneswar . If any type of requirement of Application Software, Website Design & Development, SEO, SMO Service, Ad Posting Service . SEO & Website Starts from 100$ onwards.

    ReplyDelete
  26. An idiot said this is an April Fools...
    That is probably because he hasn't got such a good brain.

    Great work Keen, keep it up!

    FYI it was that paid-by-Valve Rock, Paper, Shotgun.

    ReplyDelete
  27. Java RMI based mostly Client Server (Java Developer "www.dev2one.com") application, within which Server can use to serves requests of remote Java purchasers for
    mathematical operations. that may permit the user for following operations, Add,Subtraction, Square, and Multiply.
    java

    ReplyDelete
  28. Have you read Superintelligence by Nick Bostrom, and if so; what is your take on his take on the challenges of the value loading problem?

    ReplyDelete
  29. That's nice. To bad your logo is bad.

    ReplyDelete