Growing up, my siblings and I were essentially banned from playing video games on monitors larger or more vibrant than a monochromatic Game Boy screen—so fearful were our parents that we would actually enjoy them. In fact, I can count on two hands the number of video games I’ve played in my entire life. So I’m probably the last to notice that conversations between players and non-playable characters (characters controlled by artificial intelligence, also known as NPCs) are rarely a video game’s strong suit.
Take Elder Scrolls V: Skyrim, a gorgeous role-playing game populated by all manner of NPCs. As NPCs go, Skyrim’s denizens qualify as brilliant. They have varied personalities and purposes. Should you attack, they’ve been programmed to run away, or fight back, or plead for mercy. They don’t take kindly to you moseying through villages with your weapons drawn, or sneaking around amidst their wares. Yet you can punch a thief to death with your bare hands while, just feet away, shopkeepers calmly plot to attract more customers: “Lower the prices by half, maybe get some jugglers.”
I know, it’s a glitch. But when NPCs aren’t bizarre, they’re boring. They say their piece and never an iota more. They cannot be goaded or implored. They don’t like you very much. “By now I’ve spent hundreds of hours playing Elder Scrolls games and never been intrigued or compelled by a single character’s emotional predicament, despite these characters’ clear intention to intrigue and compel me,” writes critic Tom Bissell. How could it be otherwise? In no serious fashion are conversations interactions between two agentive intelligences. What you yourself bring to the exchange and need to take away from it have little bearing on the exchange itself, making the NPCs little more than animated information receptacles to be peered into, overturned, or bypassed altogether.
What’s missing is common ground. In linguistics parlance, common ground is a set of assumptions we share that makes communication effective. Certainly context contributes. Saying “I’m cold” at the office means something different (“Can we up the thermostat?”) than saying “I’m cold” on a camping trip (“Who’s idea was this?”) or in the middle of the night (“Stop hogging the sheets!”). Understanding who we are, the urgency in our voice, and our relationship to our listeners also affects how our proclamation is received.
But common ground is more than just context. It’s what I know you know I know, and as the world around us changes, it must be constantly re-established. If conversations relay information, common ground keeps the baton from dropping. To this end, we repeat and we question. We glean meaning from each other’s faces. We put out feelers, wanting always to know: Capiche?
Most of my gamer friends claim not to be bothered by the unresponsiveness of NPCs. “I don’t think the incongruity is too jarring for anyone who’s been playing games for their whole lives,” one friend told me. Due to technology’s limitations, generations of gamers have been conditioned to expect so little in the way of appropriate responses from NPCs that when these characters do respond appropriately—an NPC calling the cops on you in Grand Theft Auto 5, say—the gesture is a pleasant surprise. Another friend pointed out that though incongruities are “funny,” they aren’t strictly illogical. Games create their own worlds, worlds with clearly defined rules. If a world has unresponsive NPCs, well that’s just how that world works.
Maybe. But as games become more visually realistically, isn’t it natural to demand they become more behaviorally realistic as well? To be sure, this task would be immense, requiring considerable advancements in natural language processing and artificial emotional intelligence. An experimental game like Façade, though limited in scope (and execution), offers a starting point: at its best, you use both natural language and body language to counsel warring couple Trip and Grace, probing the limits of their exasperating, if nonhuman, personalities.
And NPCs don’t need to seem human. As scholar Steven Jones put its, the boundary between human and artificial actors “is not something one seeks to dissolve; on the contrary it’s the primary source of interest and pleasure.” But while we may never desire humanity from NPCs, we might nonetheless desire them to use their narrow range of game-given talents to listen and respond to us—if not to share a moment as we wash the thief’s blood from our hands, at least to hand us a towel.
