Thinking machines go pop

I take the view that computer languages, robot ethics, method acting, and biomechanics are the main ingredients that fused, in the 1950s, to become the cultural meme we call Artificial Intelligence, or AI for short. None of those ingredients was wholly new at that time: biomechanics and the Method – first known as the Stanislavski system – had been around since the early 1920s; Isaac Asimov, in conjunction with science fiction author and editor John W. Campbell, formulated the Laws of Robotics in 1940 (t’was about time, too – Jaques de Vaucanson had had the first mecha working in 1737); Ada Lovelace had anticipated the development of computer software, artificial intelligence and computer music back in 1843. But in the 1950s, thinking machines went pop. —Dirk ScherungThinking machines go pop (Robot Soul)

I’ve had only the most basic training in acting, and no formal training in artificial intelligence. It’s been productive in the classroom to apply what I do know about those topics to certain works of literature (such as Galatea 2.2, PICK UP AXE, or R.U.R.). But I’m very interested in what Scherung might find as he continues to explore this meme. Hurrah for yet another bridge across the cultural divide.

I’m particularly puzzled by the suggestion that the man pretending to be a woman in a Turing test is drawing on the same store of creativity that a method actor would use. A method actor draws on his or her own specific personal memories in order to find emotional depth that fills out the spaces in between the words the playwright wrote about the character. I can see how the attention to the construction of a character contributes to the spread of the AI meme, but I don’t know that method acting contains any truths that would be useful to the AI community.

As a homosexual, perhaps Turing was able to draw on his personal experience of gender roles to concoct the gender-bending experiment. But how does this relate to method acting? The specialized acting skills of the drag queen are campy and farcical, not offering the sort of psychological depth and individuality associated with the plays written for method acting.

You need a certain kind of physical space for method acting, and only certain kinds of plays lend themselves to method acting. The tastes of the playgoing public, the talents and accomplishments of playwrights, and the performing styles of actors are all interconnected.

Drama expresses universal themes, but it does so through unique, individual characters. I’ve raised this topic on this blog before (and when I did, I don’t think I convinced Will). But here goes… Computer programming in general is about abstraction. A program that accurately simulates the actions of a man pretending to be a woman would probably have more hard-coded, specialized features than a program that could accurately simulate general human behavior. But isn’t it specific human actions, in specific contexts, that make dramatic interest? Is human behavior, taken in general, ever that dramatic?

If I weren’t sick, and I had the time, I’d check to see how much of this has been covered by Brenda Laurel, or by Michael Mateas and Andrew Stern.

Oh, and of course I’d suggest that Rossum’s Robots be added to the list of artificial intelligence precursors. RUR was tremendously popular in its time. It also popularized the word “robot” in languages around the world.

7 thoughts on “Thinking machines go pop

  1. I can see that it’s useful to have a term to mean “the character who gets the most lines” or, in a movie, “the character who gets the most screen time,” but Scout is the narrator.

    Consider also The Great Gatsby, which is narrated by Nick, who has his own plot arc, but whose own plot is a subplot within the story of Jay Gatsby. Or Faulkner’s The Sound and the Fury, which which has three different first-person limited narrators and one third-person omniscient narrator.

    The orignal Star Trek did feature several episodes with Kirk fighting a double, but the one you’re probably thinking of is “The Enemy Within,” in which the transporter split Kirk up into a good Kirk and a bad Kirk. The good Kirk lost his ability to make decisions and be a leader, while the bad Kirk lost his ability to control his impulses.

    I’ve come to realize that even the science-fiction fans among my students have only a passing familiarity with classic Star Trek. I can imagine completely reworking those examples someday. But my main goal is to get students away from describing a car crash or a big game when they’re supposed to write a narrative essay for homework. I don’t teach any creative writing courses, so I leave the gory details to my colleagues… but your “Objective Story” and “Subjective Story” distinction seems useful.

    The structure of the original Star Trek shows made Kirk the central figure almost all the time, but the later versions of the show employed the ensemble cast. The climax of the objective story and the subjective story doesn’t always happen on the bridge where the action is. I remember an episode of Voyager where Tuvok went behind Captain Janeway’s back, making a deal with some planetary body, doing something like trading information for technology. Anyway, after the crisis is averted, the real climax of the episode came when Janeway confronted Tuvok in her office.

    Blinking lights and shaking cameras (and, in later shows, better special effects) all help to establish urgency that ratchets up the tension, thus giving the actors something to react to, and thus helping the dramatic performance. But when the climax is the special effects (whether cheesey ones, like shaking cameras, or big-bugdet explosions) you get spectacle, which Aristotle said was one of the components of drama, but far from the most important. A video game puts spectacle much higher up in the hierarchy, and that’s fine… in some genres, spectacle is going to be higher than in others.

  2. One more remark about “Crisis vs. Conflict”: the claim “a good story hinges on the will of the protagonist” is a bit misleading. The Main Character’s will determines the outcome of the story, but while it’s true that authors often make the Main Character the Protagonist, this is nothing more than a convention, probably supported by “expectations of the market”. Students should know that they forego many interesting dramatic possibilities by not considering other character constellations. Example: in “To Kill A Mockingbird”, the Protagonist is Atticus Finch, and the Antagonist is Bob Ewell. The Main Character, however, is Finch’s daughter Scout, and the Obstacle Character is Boo Radley.

  3. If I wouldn’t consider “the venting of expletives” to be “dramatic behavior”, I’d get a problem: how should my virtual character react to that? After all, it’s a strong expression of emotion, it is usually an indication of some drama taking place, and it is often the cause for the development of more drama. Plus, integrating this behavior is relatively easy, because the signal is so strong, so it’s a low-hanging fruit, it lets me define a behavior that is eminently plausible, and I can use this as a foundation to motivate other behaviors. I’m building bricks to let the player/client build houses, and those houses have to rest on a foundation. And I see the place for this foundation not in some theoretic equilibrium of dramatic forces, but in the most extreme forms of conflict that I can possibly capture.

    I’m very much aware of the differences between my approach and the approaches that most IF writers have taken so far. I’m not all that comfortable with riffing on those differences now. I find it unpleasant enough having to aggravate the engineers and scientists of AI to make my point; I don’t want to start more fires than I must. However, I’m trying to answer any specific questions as they arrive.

    Which brings me to the newsgroup post you pointed to, and to your “crisis vs. conflict” handout. Brunetière’s notion sure is sound; “crisis vs. conflict” is just another way to express the “Objective Story vs. Subjective Story” idea. Your “Captain Kirk” example, however, seems to lack some crucial information: the crystal demon and the Romulans are Kirk’s opponents in the “Objective Story”, or “crisis” – but you have to assign an opponent for the “conflict”, or “Subjective Story”, to make the example work as desired.

    In the many Star Trek episodes where Kirk is assigned to the Main Character function, most often either Spock or McCoy – depending on the specifics of the emotional content – is assigned to the Obstacle Character function (although I seem to recall one episode where Kirk is physically doubled, and we get “Good Kirk vs. Bad Kirk” duking it out).

    The full story would probably play something like this: Kirk realizes that the crystal demon must be stopped, and is ready to use the sound waves. Spock clearly backs him, because it’s “logical” – if they don’t do it, the Romulans will come, and they will use those sound waves anyway. McCoy is absolutely against it – he wants to save those eardrums at all cost. Time is running out, the conflict escalates, McCoy threatens to resign. Under pressure, Kirk comes up with a plan that would put the Enterprise at risk in a maneuver which is supposed to train the crystal demon on the Romulan cruiser. Spock opposes that, because the probability of success is very low; in his mind, Kirk just puts the ship at stake in a futile effort. Kirk does it anyway – “Scotty, give me more!” -, and, wonder o’wonders, it works – the demon and the Romulan cruiser destroy each other, the city is saved, McCoy looks Kirk in the eye and says “Thank you”, and Kirk just nods.

    So I think you need to give the full setup to make this “crisis vs. conflict” thing really comprehensible.

  4. Every intellectual undertaking has to have boundaries of some sort, since we all have to start somewhere. I didn’t notice that you had a date cutoff. In my dissertation, I chose 1950 as a cutoff date, figuring that after that time, lots had been written and lots was known about the relationship between technology and literature, but that my particular theme was under-researched in an under-researched area (American Drama from 1920-1950).

    Some of the “dramatic behavior” you describe is merely the venting of expletives (and believe me, I’ve done that). I’m not sure I’d call that dramatic — though I can understand how the word might apply in a very general sense.

    A few years ago, I asked a similar question on the interactive fiction newsgroup You might find the thread there worth perusing: “Drama in IF: Limited?

    The handout I mention in that posting has moved here:

  5. Damn, I flunked the climax, where the bridge of the Enterprise is shaken by enormous trepidations, all the lights flicker like crazy, Kirk bobs up and down in his armchair helplessly, Uhura is flung over his head and lands squarly on Chekov’s back, and the computer is launching indoor fireworks.

    Maybe I should try writing Interactive FanFic.

  6. Sorry for taking so much time before coming back to this:

    Computer programming in general is about abstraction. A program that accurately simulates the actions of a man pretending to be a woman would probably have more hard-coded, specialized features than a program that could accurately simulate general human behavior. But isn’t it specific human actions, in specific contexts, that make dramatic interest? Is human behavior, taken in general, ever that dramatic?

    If you cast a human character against a virtual character, lots of dramatic behavior is observable on the side of the human character (“Why don’t you print that, godammit!”; “Die, motherfucker!”; “Pwned!”). Not a lot is observable on the side of the virtual character. My mission is to change this – reduce this sort of behavior on the human character’s side, and increase it on the virtual character’s side.

    So the human does provide specific actions in specific contexts already – the task is to create a system where these actions and contexts are translated into dramatic values, and a virtual character can react to them appropriately.

    And re: Rossum’s Robots: of course they belong on a list of AI precursors, but since Jorn Barger’s Timeline of Knowledge Representation is available, I felt that I wouldn’t need to provide one (though I should link to it). I define the year of the publication of Turing’s idea of an Imitation Game as my cutoff point, and concentrate on the period from 1950 to the present. This is arbitrary, but no more arbitrary than choosing any other date on Barger’s list except for his very first entry:

    13,700,000,000 BC: universe expands according to human-comprehensible laws…

    And that’s not the scope that I want to try and cover with my blog :-)

Leave a Reply

Your email address will not be published. Required fields are marked *