Artificial Intelligence and the Novel

A couple of years ago, I began writing a piece for my blog about Amazon and self-publishing. I thought I had a clear idea until I started writing it. Finally, I put it aside because it seemed unclear.

Yesterday, my son Edward, a writer himself (though of music) sent me an interesting article about artificial intelligence and writing, The Great Fiction of AI by Josh Dzieza, in theverge.com

The article discusses how artificial intelligence is approaching the point at which it can write fiction. It quotes Mark McGurl in his book Everything and Less, who captured in perfect economy of words what I had been trying to say in my earlier attempt: “the Kindle platform transformed the author-reader relationship into one of service provider and customer.”

In the opening, the article shows a writer who writes a book every four months, following a project management approach. She explains that she “allots herself precisely 49 days to write and self-edit a book. This pace, she said, is just on the cusp of being unsustainably slow. She once surveyed her mailing list to ask how long readers would wait between books before abandoning her for another writer. The average was four months.”

This is dramatically different from the writer telling the story in his head and heart and (if it is good enough) giving it to the world through a publishing industry the begins with an agent, passes through editors residing in comfortable offices in New York publishing houses, through distribution, to independent booksellers (or the only big one left, Barnes and Noble), on whom the reader depends to suggest a good book.

The article details the latest efforts of Sudowrite, an AI designed for writing. Sudowrite accomplishes fiction by massive statistical analysis. Such fiction would be what that comfortably ensconced editor might look down her patrician nose at and pronounce to be “formula fiction.”

Then there’s that other stuff. The story that takes a half-year to draft and a year or more to revise. It’s often formula, too, simply because the natural progression of a story is a formula: beginning, rising action, climax, falling action, denouement. Aristotle codified it (along with so much else in western thought), but it holds through many cultures and a great deal of music and poetry as well as writing.

Presumably, AI will get better and better. After all, Deep Blue, the chess Deep Blue Chess Computersupercomputer, did finally beat a grand master. But will statistical observation allow Sudowrite to write good fiction? Hard to tell. The bluesman says, “You gotta suffer if you want to sing the blues.” If you want to write deep emotions, don’t you have to have felt them? Or can you rely on the descriptions of others to do it for you, as sampled by AI?

Not sure of the answer, but the transformation wrought by Amazon, in my mind, establishes a bifurcated world of fiction. Right now, there’s a creative wall between the fast written, formula driven novel that is being reeled in by Sudowrite and the traditional writer-in-a-garret novel.

Next problem for me: I’m on the Indie/Amazon side of distribution world and the writer-in-a-garret side of the creative wall. And that wall I mentioned may be a chasm.

But I write for passion and pleasure, and to enjoy the infinite complexity of the mother tongue, so I ain’t quittin’.

Projecting the future (everyday technology)

In the last post, I covered the two major ideas that underlie my three, going on four, novels.  I would love to have readers’ thinking on future technology (and, if you want to get into the hard stuff, social relations).

Here are some of the projections I made about everyday technology/living:

A 128-megabyte disk drive, late ’70’s

e-pad: Personal computers are becoming more and more portable.  So it was easy to envision the e-pad (no lawsuits here!), a handheld device that does computation, audio, video and so on.  Also, in Fatal Score, I assume a 3 terabyte storage device the size of a bouillon cube. I had pretty good history to go on there. Moore’s Law, which posits a doubling of computing capacity every 18 months (sometimes two years) gets me there.  After all, a 128 mb of a high-tech disk drive in the

My stick drive

early ’80’s was the size of a small refrigerator.  I just bought a 64 gigabyte stick drive (500 times the capacity of that ’80’s disk drive) for $14.95 to back up Fatal Score audio files. The three-terabyte bouillon cube in Fatal Score (23,000 times the capacity of the ’80’s disk drive) was a linear projection of Moore’s Law. (I did set the story a bit farther in the future in its original version.  Then along came the 2016 election, Russian hacking and social media attacks, and … I pulled back.)

Partially self-driving cars?  Sure, we’re already seeing the first inklings. I forecasted that we will see use of autopilot on big, well-defined roads.  So, in Fatal Score, Joe can turn on autopilot going north out of New Orleans on Highway 61, but he’s got to turn it off on the dirt roads around Panacea.  Perhaps the slowest-developing of the various technologies influencing (invading?) our lives is the vaunted Artificial Intelligence, which can now project that we’re interested in shoes … wait for it! … if we search the web for shoes.  I think it’ll take longer than the decade between now and Fatal Score go to full autodrive.

Bluetooth implants:  They enter my fiction in Skins and Bone (book 2), when the Wall Street hot shots are using them.  Joe gets one by partway through Skins. It is removed (painfully) in Fail Deadly (book 3).  The evolution of technology makes implanting make sense to me. Go to any mall (please, before Amazon wipes them out) and check out how many people have earphones in as they walk around.  How much more convenient to have the audio delivered directly to your ear through your mastoid bone.

Solar power: The plot of Fail Deadly (book 3) turns on solar power.  When I began outlining the story, solar power was not efficient enough to be truly cost competitive with traditional coal/gas/nuclear power.  Something to do with the Shockley–Queisser limit.  But there’s an often-repeated pattern in technology development that drives efficiency up (and cost down) once the science is established.  That is happening faster than I expected … the SQ Limit is being overcome by layering of transistor junctions, more efficient impedance-matching and so on. I’d better get #3 published before solar takes over!

Trains (and transportation in general): So far, I’m missing on this one.  In 2011, I projected that USA would copy Europe and (particularly) China in rapid development of rail and light-rail. I thought air travel would decline.  In my lifetime, it has fallen from exciting to necessary to grubby, and I projected that it would follow the cost curve down to disgusting. In all the books, Joe and Weezy usually ride trains or use autocars (self-driving Ubers) in cities. Instead, New York’s century-old subway system is crumbling, as is much of our infrastructure (though that’s hard for me to square with the many road construction projects in the Twin Cities).

Drones:  They are ubiquitous, though usable only by license, except for the government. Seems obvious to me that something with so much potential for danger and capability to invade private spaces will eventually be regulated.  But I guess I’m a logical optimist on that subject, given our experience to date on regulating firearms. 

So, what do you project?