Artificial Intelligence and Baby Machines of Turing
I had never read a single line Alan Turing, the father of theoretical computer science and artificial intelligence, wrote until last Sunday. I assumed it would be boring. I read his classic 1950 paper on artificial intelligence and the depth of imagination mobilized my inspiration again. There was no limit, even telepathy was worthy of discussion.
I had heard Alan Turing’s name countless times during my college days. Mostly in computer classes, especially in theoretical ones. His name would also pop up in artificial intelligence classes. The experience that followed his mention was not the most colorful one. You can picture any stereotypical college math class and you would not be that far off. It would usually go like this: First, there would first be a brief explanation of one of his ideas. Then, our professor would say he was a genius. The rest of the class would be consumed by long mathematical constructions, proofs and discussions that followed. Although I enjoyed math, I found this experience to be very monotonic. The idea, rewriting the idea in mathematical form, and then proof. Off we go!
The experience I have had with Turing then made me think that his life also mirrored our monotonic lessons. I knew his ideas, learned his math. That was already enough Turing for me. Reading his papers was definitely not on my list of fun things to do on a Sunday afternoon.
My assumptions were proven wrong last week. I was watching a video of Noam Chomsky talking about artificial intelligence. Chomsky argued that Turing thought “the question whether a machine can think is too meaningless to deserve discussion”. He explained that this core belief was why Turing had suggested his famous Turing test.
This was news to me. I had heard of the Turing test but did not know how Turing came to suggest it. The Turing test is one of the most heavily referenced concepts in computer science. The test itself simply describes a method of identifying whether a machine is intelligent. Still, it entertains an unprecedented celebrity status. It has become such a fundamental notion in academia that it is hard to find even an introductory computer science textbook which does not mention it. Not only that, its fame has reached far beyond university circles. If you find yourself having a beer or two with your software friends, you will surely score points when you bring up the Turing test.
Chomsky’s words suddenly reformed my assumptions about Turing. From a math monolith, he morphed into a normal human being, someone who was also curious about random weird questions like whether machines could think.
If Chomsky had read his paper, why not follow suit? I decided to give Turing’s classical “Computing Machinery and Intelligence” article a shot. He wrote this paper in 1950 and in one fell swoop started a whole new research field of “artificial intelligence”. It is “the paper”, that inspired schools of thinkers, mathematicians and scientists, to work towards making intelligent machines.
Before I knew it, I was reading Turing’s words. He starts his paper with this sentence:
”I propose to consider the question, “Can machines think?”
It took me the whole Sunday afternoon to read the paper.
I was stunned by what I read. I was totally stunned, taken away, flabbergasted. I was speechless. The very paper which started the whole artificial intelligence field had everything in it that we are not so used to seeing in today’s science articles. From “telepathy” to “machine education”, from “feelings of intellectual people” to “child machines”, from “Helen Keller” to “Constitution of United States”, Turing wrote about everything. He made crude estimations, asked many questions and left most unanswered. This was “the paper”, and it earnestly mused about the possibility of a human judge telepathically figuring out whether he was talking to a computer. Turing advised that his famous test should then be performed in “telepathy-proof” rooms.
As ridiculous as it was, this paper made me excited. I saw the human factor in all those gray college memories of proof hunting in my classes. I saw how brave Turing was to write about anything he imagined. He asked the fun questions. Anything was possible. Why not? He probably did not care. He was just curious.
The central question of the paper is whether we could ever build intelligent machines. Turing advocates that this is a meaningless discussion. Instead, Turing says, we should focus on writing computer programs that are good enough to trick humans believing that they are talking to another fellow human being, as in his famous Turing test. Predicting this can be done within 50 years, he ruminates on the religious, scientific and mathematical objections to this endeavor. He wraps up his paper by suggesting that instead of writing a full scale program that replicates an adult human’s cognitive ability, it would be easier to program a child machine and teach it in school like a human child to make it “intelligent”.
Turing was not boring, he was in fact quite entertaining. He liberally talked about wacky ideas, possibly too wacky for his time, and yet just wacky enough for 2010s. My love for computer science and artificial intelligence was set ablaze once again. If artificial intelligence field started with this paper, anything is possible!
Turing was a dreamer, an inch short from being a magician. And that reinforced my pride in being a software engineer. Our field is magical, and our fathers are dreamers. All we can hope for is to continue to spread this magic!
May your next wacky idea start a new field in computer science…
2) Brains, Minds and Machines Symposium, May 2011, MIT: Noam Chomsky’s talk on “artificial intelligence”
Originally published at www.onehundredyearsofcode.com.