Checkmate to Debate to S G

Instead of the checkmate game, we would prefer the debate game. Sit Selmer down across from Deep Debate, throw out a topic (how 'bout ``Is cognition computation?"), and let's go at it. When Selmer ``senses a new kind of intelligence across the table" in such a fight, well, then there may be something to write home about. We could of course ask the audience what they sense, if anything. We expect that they will be saying ``Nada" for decades to come.

Maybe the debate game is too tough. (To draw an opponent with a fighting chance, perhaps we can let a human proponent of Strong AI oppose Bringsjord.) After all, the debate game is essentially a form of the Turing Test, and though we are quite sure that a reasonably parameterized version of TT will be passed by an AI of the future, the advent of such an AI won't come in the near future. So here's an easier game. Russell and Norvig, in their excellent Artificial Intelligence: A Modern Approach [Russell and Norvig, 1995], which one of us (Bringsjord) uses to teach AI, take an approach that is now familiar to nearly all: the ``agent approach." The beauty of this approach is that it unites a field that otherwise looks disturbingly disparate -- but the approach also provides the substrate for games that go beyond classic strategy games of the sort so popular at AAAI and IJCAI. In AIMA, an agent is a mapping from percepts to behavior (see Figure 1). So let's build an agent to play the ``Short Short Story Game" (S G): The percept to the artificial player in S G is one relatively simple sentence, say: ``Barnes kept the image to himself, kept the horror locked away as best he could." (For a much better one, see the ``loaded" sentence shown in Figure 2. When will a machine give the Kafkas of this world a run for their money?) The same percept is given to the human player. Both must now fashion a short short story designed to be truly interesting; the more literary virtue, the better. The behavior in question, then, is simply producing the story (the length dimensions of which are specified, etc.) It seems to us that chessbots are arguably passé. So why not move to S G, or something similar, as the next frontier?

 
Figure 1: Russell and Norvig's Agent Scheme

There are some rather deep reasons for moving from chess (and its cognates) to something like S G. Here are three:

1. Many cognitive scientists plausibly hold that narrative is at the very heart of human cognition. For example, in their lead target chapter in Knowledge and Memory: The Real Story [Wyer, 1995], Roger Schank and Robert Abelson boldly assert on the first page that ``virtually all human knowledge" is based on stories.
2. S G strikes right at the heart of the distinction between ``Weak" and ``Strong" AI. Humans find it impossible to produce literature without adopting the points of view of characters; hence human authors generate stories by capitalizing on the fact that they are conscious in the fullest sense of the word. Ibsen, for example, described in considerable detail how he couldn't write without feeling what it was like to be one of his characters. (We return to the notion of so-called ``what it's like" consciousness below.) Chess, on the other hand, can clearly be played, and played very well, without the ``Weak" vs. ``Strong" split being touched. [For more on this second point, see [Bringsjord and Ferrucci, 1997].]
3. Despite the fact that our world is now populated with robots, softbots, immobots, and so on; despite the fact that AI continues to ascend -- there remains a question, one that is on the minds of many of those who see our progress: namely: What about creativity? As many readers will know, Lady Lovelace famously pressed against Alan Turing and his ``Turing Test" a short but powerful argument; charitably paraphrased, it runs as follows. ``Computers can't create anything. For creation requires, minimally, originating something. But computers originate nothing; they merely do that which we order them, via programs, to do" [Turing, 1964]. This argument seems to have bite against those who tout progress in checkers, chess, go and so on. It would seem likely to lose much of its force against a bot good enough to genuinely compete in S G.

 
Figure 2: S G

How do machines fare in S G? How will they fare? Bringsjord may be in a good position to ponder such questions. With help from the Luce Foundation, Apple Computer, IBM, and the NSF, he has spent the past seven years working (along with a number of others, most prominently Dave Porush, Dave Ferrucci and Marie Meteer) to build a formidable artificial storyteller. The most recent result of this toil is the agent BRUTUS , soon to debut in conjunction with the publishing of Artificial Intelligence and Literary Creativity: The State of the Art [Bringsjord and Ferrucci, 1997] from Lawrence Erlbaum. BRUTUS is a rather interesting agent; he is capable of writing short short stories like the following.

Dave Striver loved the university. He loved its ivy-covered clocktowers, its ancient and sturdy brick, and its sun-splashed verdant greens and eager youth. He also loved the fact that the university is free of the stark unforgiving trials of the business world -- only this isn't a fact: academia has its own tests, and some are as merciless as any in the marketplace. A prime example is the dissertation defense: to earn the PhD, to become a doctor, one must pass an oral examination on one's dissertation. This was a test Professor Edward Hart enjoyed giving.

Dave wanted desperately to be a doctor. But he needed the signatures of three people on the first page of his dissertation, the priceless inscriptions which, together, would certify that he had passed his defense. One of the signatures had to come from Professor Hart, and Hart had often said -- to others and to himself -- that he was honored to help Dave secure his well-earned dream.

Well before the defense, Striver gave Hart a penultimate copy of his thesis. Hart read it and told Dave that it was absolutely first-rate, and that he would gladly sign it at the defense. They even shook hands in Hart's's book-lined office. Dave noticed that Hart's's eyes were bright and trustful, and his bearing paternal.

At the defense, Dave thought that he eloquently summarized Chapter 3 of his dissertation. There were two questions, one from Professor Rogers and one from Dr. Meteer; Dave answered both, apparently to everyone's satisfaction. There were no further objections.

Professor Rogers signed. He slid the tome to Meteer; she too signed, and then slid it in front of Hart. Hart didn't move.

``Ed?" Rogers said.

Hart still sat motionless. Dave felt slightly dizzy.

``Edward, are you going to sign?"

Later, Hart sat alone in his office, in his big leather chair, saddened by Dave's failure. He tried to think of ways he could help Dave achieve his dream.

But such near-belletristic feats are possible for BRUTUS only because he (we use `he' rather than `it' in order to remain sensitive to BRUTUS 's intimate relationship to the late, corporeal Brutus, who was of course male) has command over a formalization of the concept of betrayal. (BRUTUS also has a quasi-formal account of self-deception, and provisional accounts of evil and voyeurism.) In order to adapt BRUTUS to play well in S G, he would certainly need to ``understand" not only betrayal, but other great literary themes as well -- unrequited love, revenge, jealousy, patricide, and so on. Though our intention is to craft a descendant BRUTUS , for some n > 1, that ``understands" all these literary concepts (and a lot more), perhaps S G is still a bit too tough. (At the workshop, Adam Lally can report on his attempt to build, from scratch, an agent capable of meaningfully playing S G.) Hence we briefly discuss a third type of game: infinite games.