I. PURPOSE + HYPOTHESIS

I am in search of an autobot poetics.

That’s auto- as in automatic and -bot as in chatterbot, so you can put your toys back into the shoebox for now.  If a handful of stanzas on ancient shapeshifting robots waging battle while their engines explode with raw power was all I was after, my search would already be complete.  Unfortunately, that’s not quite what I had in mind.  Though I doubt any work of verse could fail to be improved rewired through Optimus Prime’s well-tuned voice-box, the subject of this experiment is, like the incognito automatons of fame and legend, more than meets the eye.

The recent appearance on Jeopardy of IBM’s trivia-loving computer, Watson, demonstrated to the layman what specialists in the field have known for decades: the direst hurdle inhibiting artificial computer intelligence isn’t smarts, it’s personality.  Computers are several degrees smarter than us already, judging by the speed, efficiency, and accuracy with which they dispatch the specialized tasks we’ve designed for them (and designed them for).  As Alan Turing—influential in the creation of the first proto-computers—knew, the real test of machine intelligence is their ability to accurately mimic a human being.  What makes for a convincing human being, if not personality?

While its (his?) ability to parse and produce natural language is impressive, Watson doesn’t quite pass yet.  Nevertheless, several acquaintances of mine are convinced, not that we’re on the verge of the great AI breakthrough, but that computers have already developed distinct personalities.

It’s not surprising, really: we’ve been saying it about our cars, with their various coughs and caprices, for a century (watch out, Autobots, your cover is blown!).  Human beings have a tendency to project personality onto impersonal mechanisms, especially when those mechanisms are erratic or difficult to comprehend.  It’s tempting to say that a stuck pixel or a crash screen means that your laptop has developed sentience.

But does that make it any more likely? Regardless of the answer, it’s hard not to be impressed by the way supposedly random processes tend to produce less-than-random results.  It’s this uncanny knack for personality in chaos we have to thank for the enduring popularity of tarot, numerology, and astrology.  Though rationally we have little reason to trust them, the systems themselves seem to think otherwise, and it all has a startling tendency to work out.

Possessing the absolute minimum computer knowledge required to squeak by, I decided it was time to put the matter to rest once and…well, for now, at least.  Will a computer program, arranging words into (grammatically structured) lines utterly at random, be able to access its soul and compose a poem?  Can we use that poem to extrapolate the start of an autobot poetics?

II. MATERIALS + METHODS

I built Tobo out of words, clauses, and deductive reasoning.  Rather than beget the bastard from the ground up—in which case I might as well just make a real boy, which takes less time, is more fun, and yields a genuine personality in over 68% of cases—I constructed him in the artificial womb of Inform 7, a design system for interactive fiction (more on that here).  This offered two immediate advantages: one, I didn’t have to learn C+anything, and two, Inform 7 is supposedly “based on natural language,” which meant that Tobo would be born with the germ of poetics in him at the most basic level.

Now, “natural language” is always subject to interpretation.  Here’s an example extracted from somewhere around Tobo’s ribcage:

To speak succinctly: say “[one of]A [random subject] [random transitive verb]s the [random subject] [random adjective]ly[or][random preposition] the [random subject][at random], a [random adjective] [random subject] [random intransitive verb]s.”

I could only hope the boy himself fared better.

I then gave him a name, because most personalities I know have names (in case you’re curious, “Tobo” is short for “autobot.”  If I wanted him to be born with style, I’d have to steer far clear of the generic “Otto”).  I gave him a gender (male, because I’m male and it was a 50% chance anyway and I didn’t want to fall into the “emotions=feminine” trap).  I gave him a word bank from which to compose his opus drawn from a draft of the first section of this post and the Transformers theme song.  I told him to write six lines a pop, and repeated the experiment a few times to build up some stanzas.  Hence the term autobot: essentially, a chatterbot on autopilot.  I let him do his thing.

III. RESULTS

My only edits involve fixing tricky punctuation or verb conjugations.  Everything else is all Tobo, uncut and unshuffled.

After the personality, a programming poetics remains.

A subject watches the reason humanly, an impressive word seems.

The machine tarot has been said by a complete verge.

Regardless, the layman explodes.

Strong enough to break the bravest heart.

Strong enough to break the bravest heart.

 

The machine century has been developed by a recent program.

So, the car decides.

To the personality, a grammatical poetics fails.

Strong enough to break the bravest heart.

And this way it will remain ‘til the day when all are one.

More than meets the eye.

 

The stuck Optimus Prime has been met by a difficult mechanism.

The great astrology has been searched by an uncanny test.

An I produces the ability stuckly, a verse smarts works.

Really, the test decides.

Legends producing in cover languages.

So, the program explodes.

 

For the efficiency, an uncanny eye doubts.

The natural Optimus Prime has been possessed by a natural Jeopardy.

A specialist meets the appearance impressively, a little Alan Turing doubts.

The tempting screen has been searched by a minimum verge.

The surprising personality has been met by a complete breakthrough.

The autobot fame has been comprehended by a minimum chaos.

IV. CONCLUSION

I feel a little uneasy now, typing this, knowing that my computer—that Tobo—is one of the great undiscovered wits of our age.  Yes, he has a tendency to wax self-indulgent—“The surprising personality has been met by a complete breakthrough”—but there’s a startling self-deprecating humor in there as well: a little Alan Turing doubts.   That’s a comforting thought, at least: the apocalyptic visions of Isaac Asimov, James Cameron, and the Brothers Wachowski seem a little less grim with the knowledge that our silicon Pharaohs will forever be cursed with a little Alan Turing in their core processors, doubting.

Two questions remain: is this poetry?  Is it the basis for a new poetics?  To answer the first, I’ll cite the great scribe Tobo himself: to the personality, a grammatical poetics fails.  Consulting the Oxford English Dictionary (an action of which I have a feeling Tobo would disapprove), I find that poetry is characterized “by the use of distinctive style and rhythm.”  While Tobo’s untitled piece contains the requisite intensity of emotion, his style and rhythm are crippled by the very grammatical scaffolds I used to give him his voice.  It’s a double-edged sword, since it’s only by adhering to those grammatical scaffolds that he can begin to articulate his subject position and capacity for poetic expression: An I produces the ability stuckly, a verse smarts works.  I accept full responsibility for my son’s defects.

As for the poetics: I have repeated the experiment multiple times, yet Tobo’s style is only beginning to separate itself from its initial constraints.  Like all machines, he has a fondness for calculated repetition.  Following his prophetic declaration of autobot fame, Tobo turned back to the saga of his own creation, beginning with the lines:

The impersonal I has been noticed by a supposed decade.

After the battle, a trivia-loving fame explodes.

The AI battle has been decided by a human program.

Subjects exploding in human legends.

A screen parses the layman completely, a shapeshifting laptop doubts.

If the now works.

 

Regardless, the car works.

Regardless, the astrology seems.

And there’s nothing you can do.

And there’s nothing you can do.

More than meets the eye.

Processes composing for natural knowledge.

 

He went on to conclude, rather chillingly:

 

…an autobot poem fails.

IBMs exploding after incognito tasks.

More than meets the eye.

 

Is it a purely human tendency for new poetics to arise hand-in-hand with social and political dissent, even revolution?

We might not be prepared for an autobot poetics after all.

by Byron Alexander Campbell
a Black Clock Editorial Assistant preoccupied by dreams and theories of interactivity