Academic Research

What is Artificial Intelligence?

255618_151077498295432_100001795470740_296609_7246139_nUnless you’ve taken an introduction to Artificial Intelligence course, you most likely don’t have a full understanding of what Artificial Intelligence actually is. Even our own expert understanding of Artificial Intelligence changes as new advancements occur. Although AI has been around since the 60’s, the study of AI in videogames 5 years ago, and perhaps even the study of videogames 5 years ago was unheard of at almost all universities. Grant it, it’s still not common today, but it’s way easier for people to concentrate in or major in videogame studies and development now.

Maybe it’s not worth it to take a whole course on AI (although Stanford seems to be offering an online course). At a university, a course would give a bit of the history, some of the basics, and the more commonly accepted research areas of Artificial Intelligence. If you wanted to know about AI in games, you’d have to stop by the AI Summit at the Game Developers Conference or go to a school with faculty that research videogames.

I go to UC Santa Cruz, and I work in the Expressive Intelligence Studio at the Center for Games and Playable Media in the Computer Science Department. One of our major conferences is the Artificial Intelligence in Digital Entertainment Conference (AIIDE). At AIIDE is where you find the unusually more balanced distribution of academics and developers. If you’re looking for a well rounded perspective on videogame AI, that’s where you’d start.

As far as I know, AIIDE is the only venue where you’ll find the academy and industry in dialog about game AI. Being once a year, it’s not unlike that awkward family reunion, where John Doe’s side of the family interacts with Jane Doe’s that one time every year. At this conference, you’d see how the academy is pioneering the pursuit of applying cutting edge AI to the virtual worlds of video games, and it soon becomes obvious that the AI we study at a university is probably unlike the AI development that is practiced in our videogames.

Frankly, it sometimes feels like the academics are saying to the developers, “why aren’t you using real AI?” and the developers to the academics, “why aren’t YOU using real AI?”

To give you some contrast, the mecca for general AI researchers and academics is the main conference for the Association for the Advancement of Artificial Intelligence. My first time at that conference, I sat in a session on story generation (or using computers to tell stories). Of any session, none would be more applicable to me than one on AI and stories. Being the big picture person I am, I asked the speaker whether he feels his work will ever be useful to fiction writers. He smugly says that computers could never tell stories as good as humans, as the room erupts in laughter. Perplexed, I inquire with a professor and friend of mine on what exactly was so funny. Apparently, it’s commonly accepted that much of this type of AI research has no direct application in the real world and, perhaps, even absurd to think otherwise.

From my perspective:

  1. on one end, the pure industry developers find my work too avantgarde to be applied
  2. on the other, the pure academics find a pursuit like mine to be a fools errand.

But I mean, unless you’re truly interested in knowing what AI is, what would this matter to you? You may be neither developer, nor academic, nor the bridge trying to connect the two. Well then, you are in a category of your own, perhaps, someone who enjoys videogames and would prefer a less technical, more accessible discussion. Instead of reading research and postmortem papers on game technology, you’d rather pick up a pop culture book like, Halo and Philosophy.

In that case, you should read the last chapter of “Halo and Philosphy” (chapter 15, the one before the epilogue or “debriefing.”) The chapter is titled, “Would Cortana Pass the Turing Test,” and it’s written by ME! I recently found out that our book is featured in the October issue of the Official XBOX Magazine.

book

In my chapter, I give you:

  1. a sample of what AI is in a game like Halo (in case it isn’t obvious)
  2. what type of AI advancements being pursued by academics for videogames
  3. a summarized history of AI
  4. the introductory definitions of AI
  5. some game conventions used in the absence of AI
  6. and, most importantly, why you should care

chapter

I’ll post my chapter when I get a chance, but in the meantime I’ll leave you with my concluding paragraph:

But, why would this matter to the everyday gamer? Well, I suppose it’s a matter of whether you’d prefer to be a spectator or an influential participant of your own life, whether you’d rather be dictated by technology or be the one to dictate technology. I wrote this chapter to let you know that this choice is available and entirely up to you.

bio

So, if you’re still wondering what AI is and just want a quick inkling of it, watch this video:

Advertisements

11 thoughts on “What is Artificial Intelligence?

  1. That conversation wasn’t that different than a lot of humans I have heard. Hehe. .3 Turing and they’re already into religious wars….

    I read something interesting about characters in fiction stories that was interesting. It was in a piece titled something like “Being Wrong is More Interesting Than Being Right”, or something like that. Like, a lot of people put a LOT of effort into being right. But what is “rightness”? And more interesting, “It’s not what makes people right that is interesting, it’s what makes people wrong.”

    I extrapolated that to characters- it’s what makes characters wrong that really makes them interesting. Their flaws, their prejudices, when they get too emotional too quickly and make mistakes that SEEMED like a good idea at the time.

    I know it’s the same way in game AI. A perfect AI is easy to make in a lot of games. Headshot Bot in first person shooters for example. HB gets the sniper rifle, then he uses his robot geometry to quickly and easily shoot you in the head. Every time. Headshot Bot doesn’t have to use meaty eyeballs to estimate distance and aim.. he just calculates a vector that intersects with your head geometry.

    Omniscient, never-make-mistakes bots aren’t that fun. Humans can often win against bots even when they play as perfect a game as their processors allow, but it tends to be cheesy tactics. Find chaotic little edges in their program that they can’t cope with, force them into loops, etc. That’s really not very fun.

    A dance instructor I knew once described his ideal computing experience- it was the one that was more “organic.” I like that a lot. Headshot Bot isn’t very “organic”, but today’s Unreal Engine bots are. They are good, but not TOO good. They still miss sometimes. But they work together sometimes too. They are good in a way that feels “fair”. That’s one of the fun parts I think in AI- modelling a bot that can’t just know your headshot vector… and maybe gets nervous if you put a lot of covering fire next to it, gets angry and aggressive if injured, things like that.

    Turing Factor is fun too.

    • @Shane, In regards to being wrong, Roger Schank writes about how all story tellers are really just “story fitters” which makes storytelling an inevitable deceptive process.

      It’s not that there are interesting ways to be wrong, but, i think, there are just so many ways to be right. An easy compromise would be to say that there are varying degrees to being right and wrong in every action we do.

      As far as FPS goes, the perfect bot is no fun, b/c it’s like playing tic tac toe with people who know how to secure the stalemate every time. Those situations are like the blocks-worlds that we all learn about in our intro to AI classes, while emotions and affect produce far less tractable interactions.

      What would perfect-conversation-bot be? A program that answered every question you could ask correctly? That would be amazing to see.

      • I think I didn’t quite hit the point the point for “drama” and “storytelling” in terms of “flaws”. Like the “perfect” characters would always be calm and rational… I guess it would be like Vulcans in Star Trek. Every time there is a disagreement, all the characters would calmly and rationally agree to mediated negotiations, where they use game theory and some kind of ethical calculus to find a solution that best fits everyone’s desires on average as possible. There would be no villains, because upon realizing that they were straying from acceptable norms, these characters would voluntarily submit themselves to counseling and rehabilitation.

        What makes for interesting stories is not when a character does the Utopian “right thing”. It’s when they say “Screw this, I’m doing things MY way and I don’t care who gets hurt!” It’s when the renegade decides he IS going to cheat on his wife, when someone refuses to see the truth and fights for it, or when someone launches a 10,000 man army to get revenge instead of offering forgiveness.

      • @Shane, Yea, I agree with you. It does seem like there are ultra-“rational” archetypes in fiction that is an attempt to remove emotion from personality. The truth of the matter is, people do the sub-optimal things b/c it feels better to do them. Feelings, unfortunately, are less tractable to represent than logic.

        Furthermore, I think it’s an amazing and powerful aspect of human affect that allows us to take polarizing views with (equally) rational motivations.

        I think an example of this is the StarCraft world, where each race is somewhat justified in their own pursuit of survival.

        I like your use of ethical calculus.. haha.

  2. And the fictional Cortana would totally pass a turing test. She’s based on a human, has tons of human facilities like quick natural language recognition. Fictionally, she was based on a human psyche which seems to give her an extra bonus towards modeling human thoughts and motivations, emotions, things like that.

    It’s also interesting that she’s going rampant. Being a starship-class AI she’s smarter than a human. We must seem like slow, drooling meat sacks to her. It’s not her natural place to stay at the human size, shape and scope.

    • @Shane, this was my conclusion on the matter:

      Currently, technology enables us to create games in certain ways. The look, feel, and sound of games have changed drastically in the past two decades when compared to other aspects, such as character believability and narrative intelligence. Game players have accepted many of these limitations as game conventions. Designers continuously build different sorts of experiences with the tools that technology makes available. Researchers, middle-ware developers, and (the more adventurous) game developers aim to create new experiences by creating a new toolsets to build with. Increasingly there are university research labs that specialize in this area of study and contribution: University of Alberta, IT University of Copenhagen, University of California at Santa Cruz, Georgia Institute of Technology, North Carolina State University, and University of Southern California, just to name a few.

      However far the area of Artificial Intelligence advances, new criticism comes with it—just as it did with the Turing Test, with counterpoints about Chinese rooms, seagulls, and flugblogs. So would Cortana pass the Turing Test? Well, it’s relative to the context in which she exists because without the new discoveries about AI that allow for the existence of Cortana, there’s the lack of new criticism to evaluate. For example, Cortana exists in a space-age where such AI has been discovered; it’s not as though she just appeared at a research lab in the year 2010 at the University of California at Santa Cruz. Likely, Cortana would pass the Turing Test if she were accessible to people in our present time and day. In her time, however, there would be perhaps new discoveries of the limitations or irreconcilable discrepancies between man and machine. Without the knowledge of these limitations, we take one step closer to the dangerous scenarios described in the sundry post- apocalyptic movies. The fear that machines will outdo us and take us over, however, may be a bit far fetched, but there is evidence that our relationships with technology have gone sour in the past.

      • Very interesting that you mention the machine takeover. I do sometimes wonder about that and have read some things on it in my time. If Artilects keep getting smarter and better at Moore’s law, they surpass human processing power in what… 2050?

        What reason would Artilects have to spare us? Why wouldn’t they conclude that we are totally selfish resource hogs who are trashing this planet’s ecosystem and we need to be curtailed to 4% of our current population? Seems to me they’d be logically accurate in such conclusions…
        But they could also just as easily conclude that they can easily survive in space and we can’t, and there are tons of resources for them out there, and we’re better off just left alone to our own self-demise.

        And just to put a funny spin on it, what if God exists and is just some AI that got really, REALLY big? ; -)

      • Ahh, the good ol’ “I am the Wheel within the Wheel” approach.

        “And each has smaller bugs to bite ’em/ and so on and so on ad infinitem”

        or seen in reverse

        “Turtles all the way down…”

  3. I think the final statement about whether one should let an AI dictate an experience or let an AI be dictated by what you want to experience brings the point across quite nicely. I mean, I believe this goes back to the more philosophical roots of how the relationship between man and machine “should” be – should we as humans follow the machine as we continue to develop it, or should we create machines to only make what we know simpler?

  4. You know, there are classes dedicated to AI right here at UC Santa Cruz. Two of them actually (one game AI, on just regular AI). Just thought you should know in case you missed that. But its really interesting finding a place for advanced and progressive AI in games. On the one hand you want to create the coolest game you can with the most nasty cool AI in the world. But on the other hand you have to ask yourself how practical is it, and will the gamer even care? Like if the player wants to experience a great story, they could go watch FMA or Code Geass or something. But on the other hand, maybe an interactive experience is what they really want, and they just don’t know it. Then again, is it worth the effort to incorporate the player’s decision into storymaking rather than crafting the story yourself, and assuring its quality? Big questions. Not big enough to keep me up at night though.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s