Are you ready for this: “Don’t miss this exciting new thriller, The Mount Rushmore Murders, by r2d2.”
It wasn’t terribly long ago that if you wanted to know something, you had to look it up in a book. Now we’re used to finding tons of information by doing a quick google search. If we’re too lazy for that, we might ask Siri, Alexa, or some other artificial intelligence (AI) creature.
In Project Debater, a computer, armed with a massive wealth of knowledge, went up against a champion debater. The computer clearly had more facts at its disposal than the human. The audience, however, declared the human as the winner, not because he won points on knowledge, but probably because he was able to make statements that were off-point yet seemingly meaningful, a tactic often used by politicians and TV pundits. If the audience had been more thoughtful, perhaps the computer would have won.
According to Moore’s law from 1965 (named after Intel co-founder Gordon E. Moore), computing power doubles every twelve to eighteen months. This sustained increase in computing power has led many people to predict that one day a computer will pass the Turing test.
The Turing test was put forward in 1950 by the computer scientist, cryptanalyst, and mathematician, Alan Turing. He proposed it as a simple way to determine if a computer is truly intelligent, i.e., able to think like a human.
The test has taken on many forms but basically requires a questioner and a responder who are hidden from each other. The questioner interrogates the responder and is subsequently asked to decide whether or not the responder is the computer. This exercise is repeated a number of times. If the questioner thinks the computer is human in at least half of the trials, that computer is considered to have passed the Turing test. Turing had predicted computers would be able to trick humans into thinking they were real by the year 2000.
At first, the test was performed using yes/no questions about a specific subject, and all input and output were typed. The game was upped when output was changed to free-form answers. Now, the test can be performed using spoken language on the part of the interrogator and even, theoretically, the answerer.
A computer named Eugene Goostman is said by some to have passed the Turing test in 2014, using unrestricted conversation. However, the computer masqueraded as a 13-year-old non-native English speaker, so the judges excused some of its poor and/or illogical communication on the basis of the respondent’s immaturity and poor grasp of English. For that reason, not all experts agree that Eugene Goostman passed the test.
One may quibble over whether the Turing test has been passed, but there is no denying that the field of AI has taken off. This has led some, like Elon Musk, to fear where it may lead. While he may be worried about robots taking over our planet, other concerns are on the horizon.
In addition to accessing an enormous amount of information, calculating, making algorithmic decisions, and dominating in jeopardy and chess, computers are making inroads in the creative arts. Since the 1980s, neural networks have been developed to predict next notes (for music) or sketch lines (for art). Already, AI is being used to create original music and art, e.g., Google’s Magenta https://magenta.tensorflow.org/.
This brings me to an area of exploration I personally find disturbing. Computer scientists are now probing AI’s ability to write creatively. Will computers successfully write interesting TV shows, movies, and novels in the future? Will we be reading books created by a distant relative of Eugene Goostman in twenty years? Will there still be a market for fiction penned by mere humans? Are youth taking writing classes in the hope of becoming successful authors wasting their time?
Before all you writers out there start smashing your computers and typewriters, be assured that such a scenario is not just around the corner. But it may be around several corners.
I recently came across an article reprinted from the Los Angeles Review of Books by Patrick House, titled I, Language Robot. In it, the reader gets a glimpse of a language bot being developed by a San Francisco AI research lab. Basically, the computer fills in words according to how likely specific words follow or precede other words (a bit like the predictive wording used in instant messages that gets me in trouble now and then). As a reference, the bot uses about 8 million documents (written by humans, I assume).
Parts of known literary works have been changed by the computer and given to literary experts. They have failed to identify the computer-generated passages. One of the literary works was from Shakespeare’s King Lear. So far, the bot is more of a writing partner than a sole author.
In a 1958 interview, Ernest Hemingway was asked about how he rewrites. He gave the following answer:
“Most of the time I just sit down and write the lines on the piece of paper. If there are any changes I make I usually go back over it and rewrite the line until I get it exact the way I want it.”
When asked about the function of his art, he answered:
“I’m afraid to answer that for fear of being laughed at. To answer that, you have to get at the heart of how a writer creates reality. It is a question I always have to ask myself: ‘Who, exactly, is doing the authoring?’ The answer to that question is usually not me. It’s the readers. It is the readers who author the work, who create the truth.”
Actually, the above are the bot’s answers. Hemingway’s answer were:
“I always rewrite each day up to the point where I stopped. When it is all finished, naturally you go over it. You get another chance to correct and rewrite. When someone else types it, and you see it clean in type. The last chance is in the proofs. You’re grateful for these different chances.”
and
“Why be puzzled by that? From things that have happened and from things as they exist and from all things that you know and all those you cannot know, you make something through your invention that is not a representation but a whole new thing truer than anything true and alive, and you make it alive, and if you make it well enough, you give it immortality. That is why you write and for no other reason that you know of. But what about all the reasons that no one knows?”
Did the bot fool you?
If it did, and you’re a writer, perhaps you should be worried. I predict it’s only a question of time before these bots start generating original ideas and plots. Most would agree that computers today cannot actually think. But what about five or ten years from now? Will we be reading novels created by r2d2, Eugene Goostman or some other computer? Stay tuned.