How backstories make artificial beings more human
Oct 25 2016
In this Halloween season, let’s talk about something scary: Artificial Intelligence (“AI”). Did you realize you have much in common with people who build robots?
“AI” is similar in ways to “AC,” a term we just coined for “Artificial Character” … in other words, any character that you’ve made up. Both can be intimidating to the people who create them and creepy to people who encounter them. But they shouldn’t be. Ultimately, AI and AC both are characters, and both become more real, more interesting – and less scary – if you take care to “humanize” them. One of the best ways to do that is to give them a backstory.
Backstories aren’t news to experienced actors, including voice actors. They’re especially common in animation (and sometimes gaming) where a characterization might be especially rich. What would your favorite superhero be without their backstory? Suppose you’re called upon to ad-lib in character – do you laugh readily, or reluctantly … and what of any other emotion?
As we noted in our article on “Newton’s Third Law of Physics as applied to Voice Acting,” an actor with no lines nevertheless reacts (and on screen or stage might even upstage the speaking actor), with “extraordinary things to say, just choosing not to say them.” Okay, in voiceover, your audience can’t see you react. But when it comes time for you to speak, where is your tone of voice coming from? Perhaps its shaded by something logical and relevant, but which only you will know. Because your character has a backstory.
This is particularly relevant when you’re developing a character all your own … say “Gerald, the Curmudgeonly Old Man,” or “Ruby the Rich Widow” or “Hortense the Overly Familiar Hog.” It not only helps you suss out character traits and vocal characteristics, but it also helps you keep the character stable in various situations and over time.
But it’s also relevant to a character you’ve had just minutes to come up with after a brief acquaintance with the script. When you break down the script, noting where the “plot” or emotion changes, how does your character react to that change?
At the least, it will help you gauge the amount and nature of non-verbal interjections you might use. Sounds that we all use, in various measures, like “Hmmm,” “Oh,” and “(chuckle).”
Little things like that, properly timed and judiciously used, so they don’t obscure your lines or anyone else’s, help humanize an “artificial” character, a character you’ve made up.
They also humanize characters who don’t emerge from your brain but instead come from an Artificial Intelligence.
AI has crept intimately close to us in the form of virtual assistants – namely, Siri, Alexa, Cortana and the like. As noted in an article in The Washington Post (“The next hot job in Silicon Valley is for poets”), newer virtual assistants are planned to have more personality than even Siri, Alexa, and Cortana — themselves a far cry from Clippy, Microsoft’s on-screen paperclip. The added personality is meant to help the new assistants relate better with the humans they are assisting.
Microsoft has already given Alexa the ability to interject non-verbal “ums” and such, and Siri is something of a jokester. Now the goal is to make these intelligences even more truly conversational.
“You have to develop an entire backstory — even if you never use it,” said Robyn Ewing, a Hollywood writer currently creating a personality for a smartphone app (called “Sophie”) that reminds patients to take their meds.
The point is to make the AI assistant both appealing and desirable. It sounds to us like a combination of marketing appeal and close human interaction. (After all, when those goals are approached well, they’re usually intertwined, aren’t they?) It also makes the character flexible – just as a backstory helps you ad-lib.
At the same time, it helps define the limits of the character’s behavior. As Microsoft learned the hard way (or maybe demonstrated?) with their Twitter bot, Tay, a learning chatbot can be induced by users to learn bad behavior, picking up and repeating inappropriate s****l or social comments if not adequately inhibited. As with real people, the character’s basic nature draws the containing lines. It is impossible for programmers to predict every possible influence and situation, but a robust personality can fill gaps and set boundaries.
Equally important is the element of too much naturalness. A phenomenon known as “the uncanny valley” occurs when a humanoid robot appears not obviously artificial, but still not quite human. To users, it can be unnerving, even scary or revolting. Just as a VO character’s voice should virtually never be tiresome or wholly obnoxious, ditto for a character’s personality unless for dramatic effect. “Uncanniness” can be both distracting and obnoxious, especially when what’s under the surface is just a little bit off. Or when what’s underneath a perfect skin is known to be imperfect.
There’s no way yet that a complex Artificial Intelligence can be totally perfect, so the developers of today’s AI, therefore, backtrack a bit — they have their wards acknowledge their fallibility. Instead of trying to appear perfectly human, they are made to seem endearingly flawed, taking their relative fallibility with a sense of humor. In turn, this makes them seem even more acceptably human. Ironic, huh? Well, not so ironic when you consider it this way: Their goal is to have the robot be a “real” robot, just as your goal is to be a real curmudgeon, a real rich widow, or a real passionate pig.
Can you achieve such subtleties in your characters? Give them backstories, and give it a try.