The Turing Test: Is that Human or Machine?

I propose to consider the following question, ‘Can machines think?’

Thus begins Alan Turing’s paper “Computing machinery and intelligence.”  It’s 1950 England, and the world’s first computer is being used to calculate the next known largest prime number, a feat meant to show off the power of the computer.  For Turing, the implications of this work go much further and spark a philosophical question: could computers one day acquire cognitive abilities rivaling human intelligence?

So, how do you go about answering this question?  Traditional approaches force you to define key terms like “thought,” “machine,” and “intelligence,” but these are vague and too hard to define.  Instead, Turing posed a slightly different quandary: “Can machines do what we, humans, as thinking entities, do?”  This simplified question allows for a distinction between the physical and intellectual properties of humans.  To demonstrate this approach, Turing proposed a test called the imitation game.  This was based on a party game in which a man and a woman go into separate rooms, and guests try to tell them apart by writing a series of questions and reading the typewritten answers sent back.  In this game both the man and the woman aim to convince guests that he or she is the other person.  How would this game work with a human and a machine?  Will the guests decide wrongly just as often as when the game is played between a man and a woman?  Turing argued that if the guests could not distinguish human from machine by questioning, then it would be unreasonable not to call the machine intelligent, since we judge other people’s intelligence based on external observations in much the same way.


Turing predicted that machines would be able to pass this test.  Indeed, just last year, a computer program, called Eugene Goostman, became the first to successfully pass a Turing Test.  And just a few months ago, researchers successfully designed a Visual Turing Test, in which the computer learned to understand the contextual relationships and implied activities between objects within photos.

Visual Turing Test

But even with these advancements, it’s difficult to know if we are any closer to true artificial intelligence.  In fact, after Eugene passed the Turing Test, critics were quick to highlight just how low the bar was set: with grammatical mistakes and poor general knowledge consistent with the bot posing as a Ukrainian teenager, Eugene didn’t actually prove that it was “intelligent.”  Now, researchers are pushing for true, real-world intelligence: a multidimensional trait that involves many skills.

Sure, computers are a lot faster than humans at mathematics, but there are plenty of ways in which computers are no match for humans at all!  For example, we can follow instructions to put together furniture with relative ease, but no bot has passed this “build-it” challenge.  Likewise, humans have common sense and intuition, but computers find it difficult to answer questions such as, “The trophy would not fit in the suitcase because it was too big.  What was too big, the trophy or the suitcase?”  Similarly, when we watch television shows and movies, we can easily answer questions that require critical thinking skills, such as ones about the motivations of characters and implications of plot twists, but robots often fail this comprehension challenge.

With the future of artificial intelligence moving away from the Turing Test in favor of more complicated and all-inclusive tests, perhaps we will get closer to the world that Turing envisioned, where machines and humans are indistinguishable.  Of course, this is many years away, and it’ll be exciting to see the advancements in this field in the forthcoming years.  As Alan Turning said,

We can only see a short distance ahead, but we can see plenty there that needs to be done.




Geman, D., Geman, S., Hallonquist, N., & Younes, L. (2015). Visual Turing test for computer vision systems. Proceedings of the National Academy of Sciences of the United States of America, 112(12), 3618–3623.

Turing, A.M. (1950). Computing machinery and intelligence. Mind, 59, 433-460.

Images made by Jooyeun Lee.

Kate Fehlhaber

Kate graduated from Scripps College in 2009 with a Bachelor of Arts degree in Neuroscience, completing the cellular and molecular track with honors. As an undergraduate, she studied long-term plasticity in models of Parkinson’s disease in a neurobiology lab at University of California, Los Angeles. She continued this research as lab manager before entering the University of Southern California Neuroscience graduate program in 2011 and then transferring to UCLA in 2013. She completed her PhD in 2017, where her research focused on understanding the communication between neurons in the eye. Kate founded Knowing Neurons in 2011, and her passion for creative science communication has continued to grow.

One thought on “The Turing Test: Is that Human or Machine?

  • August 19, 2015 at 12:49 pm

    Nice article! I don’t think that we are any closer to a real generally-intelligent machine. Instead, it looks like economics is driving AI into specific task-oriented intelligence (by which I mean flexible and robust in noisy environments).

    Are there any researchers working on general AI?

Comments are closed.