Marks: Artificial intelligence is no more creative than a pencil
Some have claimed that AI is creative. But “creativity” is a vague term. To talk about creativity fruitfully, the term must be defined in such a way that everyone is talking about the same thing and no one is changing the meaning to suit their purpose. In this chapter and the following, we will explore what creativity is, and in the end it will become clear that, properly defined, AI is no more creative than a pencil.
Creativity: Create something new
Lady Ada Lovelace (1815–1852), daughter of poet George Gordon, Lord Byron, was the first computer programmer to write algorithms for a machine that was planned but never built.1 She was also probably the first to note that computers will not be creative, that is, they cannot create something new. She wrote in 1842 that the computer “has no claim to create anything. It can do [only] all we know how to order it to operate.2
Alan Turing disagreed. Turing is often called the father of computing, having pioneered the idea of modern computers in the 1930s.3 Turing argued that we can’t even be sure that humans create, because humans don’t do “nothing new under the sun” – but they surprise us. Likewise, he said, “the machines surprise me very frequently”. So maybe, he argued, it’s the element of surprise that’s relevant, not the ability to create something new.4
Machines can surprise us if they are programmed by humans to surprise us, or if the programmer made a mistake and therefore experienced an unexpected result. 5 Often, however, the surprise comes as a result of the successful implementation of computational research that explores a myriad of solutions to a problem. The solution chosen by the computer may be unexpected. Computer code that searches among different solutions, however, is not creative. The credit for the creativity belongs to the computer programmer who chose the set of solutions to explore. Shortly we will give examples of computer searches for making the best move in the game of GO and for simulated swarms. Both results are surprising and unexpected, but there is no creativity provided by the computer code.
The flawed Turing test
Alan Turing, an atheist, wanted to show that we are machines and that computers could be creative. Turing equated intelligence with problem solving, did not consider matters of consciousness and emotion,6 and called people “human computers”.seven Turing’s version of the “imitation game” was proposed to show that computers could duplicate the conversational human. That’s why the biopic starring Benedict Cumberbatch as Turing was called The Imitation Game.
How can computers imitate humans, according to Turing?
The imitation game (which has been called the Turing test) simply asks whether, in a conversational exchange using text (i.e. an exchange in which the participants are hidden from each other), a sufficiently sophisticated computer can be distinguished from a human being. . If an interlocutor gets lucid, human-sounding responses from the computer and believes that the computer is actually a human typing responses from another room, then the test has been passed.
(Incidentally, the inverse of the Turing test is easy. Just ask it to calculate the cube root of twelve to ten significant digits. If the answer is almost immediate, you’re talking to a computer.)
There are those who claim that today’s chatbots8 pass the Turing test. Even more famously, a computer program known as “Eugene Goostman” is said to have been adopted. A caption posted claiming it reads: “‘Eugene Goostman’ tricks 33% of questioners into thinking he’s human, in what’s seen as a milestone in artificial intelligence.”9
But those who take a test often cheat to alter the result. That’s what Goostman’s programmers did. Goostman was supposed to be a Ukrainian, which meant that English was not his first language, so his bad grammar could be excused. He was supposed to be a teenager, so any lack of deep understanding could be attributed to his naïve intellect. Likewise, his tendency to be silly and deflect questions could be considered typical adolescent behavior. In short, Goostman’s software character was played to pass the Turing test.
Here are some examples of an exchange between Goostman and his interlocutor.ten Goostman is first asked a question offered in the first line of the lyrics of the song “Close to You”, popularized by the Carpenters:
Questioner: Why do the birds suddenly appear?
Goostman: Just because 2 plus 2 is 5! By the way, what is your profession? I mean, could you tell me about your work?
Here we see a cheat often used in chatbots. If an answer is not known, the chatbot responds with an unrelated question to derail the direction of the conversation.
Here is another exchange with Eugene Goostman:
Questioner: It’s okay, I’m sick of sick people. How’s your stomach today? Is he upset perhaps?
Goostman: I don’t think you can smile at all. I bet you work at a funeral agency.
As you can see, Goostman’s answers here are elusive.
Selmer Bringsjord correctly notes that the Turing test is manipulated by programmers. “Gamed” here is a nice word for being an elusive cheater. As Bringsjord writes, “Although progress towards Turing’s dream is being made, it comes only through clever but superficial trickery.”11
When playing with the system, chatbots can deflect detection by answering questions with other questions, giving evasive answers, or admitting ignorance. They display a general intellectual superficiality with regard to creativity and depth of understanding.
Goostman answered questions such as “By the way, what is your profession?” He also tried to change the subject with conversational responses such as “I bet you work at a funeral home”. These are examples of the “smart but superficial trickery” criticized by Bringsjord.
So what do the Turing tests prove? Only these clever programmers can trick gullible or uninitiated people into believing that they are interacting with a human. Confusing something with human does not make it human. Programming to superficially imitate thought is not the same as thinking. Disjointed randomness (like the topic-changing questions Goostman spat out) doesn’t show creativity.
“I propose to examine the question: ‘Can machines think?’ Turing said. Ironically, Turing not only failed in his attempt to show that machines can be conversationally creative, but also developed computer science that shows humans are not calculable.
You can also read previous excerpts posted here:
Why you are not – and cannot be – calculable. A computer science professor explains in a new book that computer intelligence has nothing to do with human intelligence. In this excerpt from his forthcoming book, Non-Computable You, Robert J. Marks shows why most human experiences are not even computable.
The Software of Gaps: An Excerpt from Non-Computable You. In his just published book, Robert J. Marks argues that consciousness is emerging from AI and that we can download our brains. It reminds us of the story of the boy who dug in a pile of manure because he was sure that…underneath all that poop there MUST be a pony!
1 Lovelace is often credited with writing an algorithm for Charles Babbage’s “Analytical Engine”, a machine that was planned but never built. There is some controversy as to whether Lovelace or Babbage wrote this first program. In any case, Lovelace was undoubtedly involved to a great extent with the very first computer programs, and she was also the first to say that a computer could be programmed to do more than just calculate. For an overview of her contributions, see Christopher Hollings, Ursula Martin and Adrian Rice, “Ada Lovelace and the analytical engine”, Bodleian Libraries (July 26, 2018).
2 Mrs Lovelace, Annex I to Faster Than Thought: A Symposium on Digital Computing Machines, ed. BV Bowden (London: Pitman, 1953), 398.
3 For an overview of the similarities and differences between Babbage’s and Turing’s machines, see Nathan Zeldes, “Babbage and Turing: Two Paths to Inventing the computer, Nathan Zeldes (website), April 29, 2021.
4 Alan Turing, “Computing Machines and Intelligence”, Mind 49, no. 235 (October 1950): 433–460.
5 Turing admits that when machines surprise him, it’s usually because of traceable human error in his calculations. It also anticipates the objection that the “surprises” of the machine are “due to a creative mental act on my part, and reflect no credit on the machine”, but does not respond to this objection, except to say that it brings back to the question of consciousness, which “we must regard as closed.” Turing, “Computing Machinery and Intelligence”, section titled “Lady Lovelace’s Objection”.
6 Turing, “Computer Machinery and Intelligence”, section titled “The Argument from Consciousness”.
seven In Turing’s “Computing Machinery” article, he refers to “human computers” no less than ten times.
8 Chatbots are computer programs that respond in a human way in textual exchanges. You may have interacted with a chat box that popped up on your screen to offer help or answer questions. Chatbots are useful for businesses because they can function as digital help desks. But they are limited in the questions they can answer and can only answer if they are programmed to answer.
9 Press Association, “A computer simulating a 13-year-old boy becomes First to pass the Turing test, Guardian, June 9, 2014.
ten George D. Montanez, “Detecting Intelligence: The Turing Test and Other Design Detection Methodologies”, Proceedings of the 8th International Conference on Agents and Artificial Intelligence, vol. 2 (Setubal, Portugal: Scientific and Technological Publications, 2016), 517–523.
11 Selmer Bringsjord, Paul Bello and David Ferrucci, “Creativity, the Turing Test and the (Better) Lovelace Test”, in The Turing Test: The Elusive Standard of Artificial Intelligence, ed. James H. Moor (Boston: Kluwer Academic Publishers, 2003), 215–239.