Can AI Write the Great American Novel? Or Compose Sports News?It’s a split decision, say Rensselaer prof Selmer Bringsjord and Baylor computer engineering prof Robert J. Marks
In a recent podcast, Rensselaer professor Selmer Bringsjord discusses AI and creativity with computer engineering professor and Walter Bradley Center director Robert J. Marks.
The difference between writing novels and playing games like Go and chess is that writing novels does not mean winning according to a set of rules. A machine can be programmed with rules and do the calculations faster—much, much faster—than a human. A good novel requires creativity in the face of situations that are only partly definable. If a novel succeeds, many people agree that the writer has captured essential elements of human nature and life circumstances. That’s what makes the great novels so memorable.
Sports reporting is somewhere in the middle in that a great proportion of it is really the regurgitation of statistics, cliches, factoids, and trivia.
Here are some excerpts from the transcript of “Bingecast: Selmer Bringsjord on the Lovelace Test”: (Other discussions in the series, Show Notes, Resources, and a link to the complete transcript follow.)
Selmer Bringsjord (pictured): However Alpha Go or any other such machine arrives at its ability to play really good Go, if there was enough time and energy available, Go is absolutely not a difficult game because we already have perfection. We can define it, so that really doesn’t count.
A novel would be rather a different affair. I just listened in audio book form to Flannery O’Connor’s The Violent Bear It Away (1960). And what struck me about this novel is that, over and over again, she makes you feel what a character is feeling with just a few sentences. So, her consciousness is present and exploited, and then she comes up with these amazing sentences, two or three in a row that just get the job done.
Note: From The Violent Bear It Away
“The world was made for the dead. Think of all the dead there are…There’s a million times more dead than living and the dead are dead a million times longer than the living are alive…”
“He kept on digging but the grave did not get any deeper. “The dead are poor,” he said in the voice of the stranger. You can’t be any poorer than dead.”– from Goodreads
Selmer Bringsjord: We can’t say that the production of sentences like that is even in theory possible by following an algorithm, because we don’t have the algorithm.
Note: Bringsjord has proposed the Lovelace test for creativity for computers. The iconic Turing test focuses on the program’s ability to fool a human. But, as Bringsjord notes, fooling a naive human might be comparatively easy without proving anything. Tests should demonstrate something beyond an individual’s ability to be fooled. See, for example, Marks and George Montañez explaining “How you can really know you are talking to a computer.
Talk turned to ways AI is now used to automate conventional copy in media:
Robert J. Marks : USA Today uses something called Wibbitz. News Tracer is used by Reuters to actually compose short reports, and BuzzBot was originally designed to crowdsource reporting from the Republican National Convention. So yeah, all of these are actually conglomerating information that already exists.
You talk about football, it strikes me that much of sports radio could be probably replicated with artificial intelligence. They say the same things over and over again like, “Frank, it’s all about getting points on the board.” “It’s going to be all about whether we can move the ball.” “We need to stop their passing game,” all of this stuff. And I bet that could be replaced by AI very easily, but that’s my takeaway.
The writing of jargon-heavy academic papers can also be automated using algorithms. The resulting nonsense often escapes detection at first. Marks tried it himself, for fun:
Robert J. Marks (pictured): Now there’s some science paper generators too, and this is really cool. One of them is called Sci-Gen. It generates computer science papers and they’re really weird, but they’ve actually been accepted for publications in places like IEEE and Springer where the peer review process wasn’t at their top. They were later found out and removed.
And another one was MathGen. You can write your own math paper. I entered our names and we have a paper now, Selmer Bringsjord and Robert Marks, called “Positive Pseudo Independent Finite Submarines and Rational Algebra.” And this great name was generated by it. But I read it and you’re right. If you go to any length, you detect that there’s something phony. Especially, the greater domain expertise you bring to the paper, the quicker you recognize it. I don’t know if you’re … do you remember the movie, The Stepford Wives, or Invasion of the Body Snatchers?
Selmer Bringsjord: Yes. I mean, not scene by scene, but certainly…
Robert J. Marks: In both of those there were robots, or in Invasion of the Body Snatchers, it was an artificially generated human via these pods from outer space. But you looked at them, they looked human and they acted human, but if you were with them for a bit, you realized that they were phony. And that’s kind of been my experience with these writers, like the MathGen papers and the Sci-Gen papers. They look good when you look at them, but you get a little bit in-depth and you realize they’re totally phony.
Perhaps the question that is coming into focus is, in relation to creativity, does lived experience matter?
You may also enjoy some of Robert J. Marks and Selmer Bringsjord,’s earlier dialogues:
Gödel and God: a surprising history: A thought-provoking account of master logician Gödel’s largely unknown proof of the existence of God.
Why our minds can’t really be uploaded to computers. The basic problem is that human minds aren’t “computable.” Peter and Jane are not bits and bytes.
Thinking machines? The Lovelace test raises the stakes. The Turing test has had a free ride in science media for far too long, says an AI expert.
Thinking machines: Has the Lovelace test been passed? Surprising results do not equate to creativity. Is there such a thing as machine creativity?
- 0:00:54 | Introducing Selmer Bringsjord, Professor — Rensselaer Polytechnic Institute (RPI)
- 0:01:49 | What is the Turing test?
- 0:04:03 | The Lovelace objection
- 0:04:29 | Ada Lovelace
- 0:07:51 | The consciousness objection
- 0:09:00 | Eugene Goostman
- 0:09:42 | The Lovelace test
- 0:13:54 | AlphaGo’s “creative” move
- 0:16:48 | Creative writing
- 0:18:17 | Has the Lovelace test been passed?
- 0:21:35 | How could it be proven that the Lovelace test was passed?
- 0:25:08 | Ray Kurzweil and singularity
- 0:26:30 | Can computers fake all human behaviors?
- 0:27:12 | Can computers duplicate all human behaviors?
- 0:28:21 | Subjective measurement
- 0:29:46 | The definition of cognition
- 0:34:01 | Is consciousness is a special case of cognition?
- 0:38:28 | Are consciousness and cognition non-algorithmic?
- 0:40:00 | Examples of cognition that are not computable
- 0:46:45 | Will AI ever write creative prose?
- 0:50:08 | The need for characters
- 0:51:17 | The mechanics of writing and Proust
- 0:53:23 | The complexity of language
- 0:56:48 | AI and financial reports
- 0:59:43 | Sports reporting
- 1:02:55 | Paper generators (Sci-Gen, MathGen)
- 1:04:27 | Recognizing AI generated content
- 1:09:37 | A book about Kurt Gödel
- 1:10:59 | Gödel’s “God Theorem”
- 1:13:06 | Was Gödel a deist?
- 1:15:37 | The history of Gödel’s proof
- 1:21:53 | The reasoning behind Anselm’s ontological proof
Gödel’s Ontological Proof
- Selmer Bringsjord’s website
- The Turing test at Encyclopædia Britannica
- Ada Lovelace at Encyclopædia Britannica
- Charles Babbage at Encyclopædia Britannica
- Bringsjord, Selmer, Paul Bello, and David Ferrucci. “Creativity, the Turing test, and the (better) Lovelace test.” The Turing Test. Springer, Dordrecht, 2003. 215-239. Link: http://kryten.mm.rpi.edu/SELPAP/REPRINTS/LOVELACE/lovelace.pdf
- Eugene Goostman at Wikipedia
- AlphaGo at Wikipedia
- Ray Kurzweil
- David Gelernter at Wikipedia
- What Robots Can and Can’t Be by Selmer Bringsjord
- Bringsjord, Selmer, and Michael Zenzen. “Cognition is not computation: The argument from irreversibility.” Synthese 113, no. 2 (1997): 285-320.
- Claude Shannon at Encyclopædia Britannica
- Bringsjord, Selmer, and David Ferrucci. Artificial Intelligence and Literary Creativity: Inside the Mind of BRUTUS, a Storytelling Machine. Psychology Press, 1999.
- Marcel Proust at Wikipedia
- Ex Machina at IMDB
- “Artificial Intelligence Can Now Write Amazing Content — What Does That Mean For Humans?” at Forbes
- “What News-Writing Bots Mean for the Future of Journalism” at WIRED
- “The First Novel Written by AI Is Here—and It’s as Weird as You’d Expect It to Be” at SingularityHub
- Generate a random Computer Science paper using SCIGen
- Generate a random Math paper using MathGen
- “Belletristic” definition at Dictionary.com
- “The Argument for God’s Existence from AI” by Selmer Bringsjord
- The Gödel solution at Wikipedia
- Gottfried Wilhelm Leibniz at Encyclopædia Britannica
- Saint Anselm of Canterbury at Stanford Encyclopedia of Philosophy
- Rene Descartes at Encyclopædia Britannica
- “On the Logic of the Ontological Argument” by Paul Oppenheimer and Edward Zalta
- Disjunctive syllogism at Wikipedia