recent movie, the Imitation Game, brought attention to the Turing Test to a general audience. First proposed by the British mathematician
Alan Turing, the test basically proposes that computers will have achieved
artificial intelligence when a person interacting with that computer cannot
distinguish between it and another human being.
year, it was reported that a machine successfully passed the Turing Test – sort
of. (See this article
in the Washington Post, for example.)
While that particular test didn’t set a very high standard, there is no
doubt that machines are getting better at doing things that humans only used to
past Sunday, there was an article in the New York Times Weekly Review titled “If
an Algorithm Wrote This, How Would You Even Know?” Its (presumably human) author warned us that
“a shocking amount of what we’re reading is created not by humans, but by
newspaper even offered its own version of the Turing Test – a test
of your ability to determine when a paragraph was written by a person or a
machine. Try it. (Disclosure: I didn’t get it right 100% of
the time, either.)
of course, it doesn’t stop with writing.
More interesting is the use of computers to be creative.
This is because
among the differences between humans and other animals, that some people claim,
is our ability to produce creative works of art. Perhaps going back to the cave paintings,
this seems to be an unusual human trait.
(Of course, you can always find someone else who will dispute this, as
in this article, “12
artsy animals that paint”.)
we may find animals painting to be amusing, perhaps we’d find machines becoming
creative as more threatening.
Simon Colton is one of the leaders in this field – which, by the way, goes back
at least two decades. He has written:
“Our position is that, if we perceive that the software has been skillful, appreciative and imaginative, then, regardless of the behaviour of the consumer or programmer, the software should be considered creative.”
He and his team have worked with software called Painting Fool. This post has some examples of that artwork, so you can judge for yourself if you could tell a computer generated it.
I have my own little twist on this story from more than ten years ago. I met some artist/businessmen who designed and then had painted high end rock concert t-shirts. These were intended to be sold at the concert in relatively small quantities, as an additional form of revenue.
The artist would prepare the design and then have other people paint them. But this was a slow tedious process so we discussed the use of robots to take over this process. (At the time, the role of robotic painting machines in auto factories was becoming well known.)
One of the businessmen posed an obstacle by noting that people bought the hand-painted product because each was slightly different, given the variation between artists who painted them and even the subtle changes of one artist on any day. I somewhat shocked him by pointing out that, yes, even that kind of randomness could be computer generated and his customers would not likely be able to tell the difference.
But, perhaps, computers could tell the difference. A computer algorithm correctly identified Jackson Pollock paintings, as reported in a recent article in the International Journal of Art and Technology. (A less technical summary of this work can be found in a Science Spot article of a few weeks ago.)
In the end, they didn’t use robots because they were too expensive compared to artists in the Philippines or wherever it was they hired them. Now, the robots are much cheaper, so maybe I should revive the idea
Anyway, we’re likely to see even more impressive works of creativity by computer software and/or by artists working with computer software. The fun is just beginning.
© 2015 Norman Jacknis