Benjamin Pineros* says a one-word Turing test has determined that the word ‘poop’ distinguishes humans from machines.
Photo: Jack Harner
Scientists have created a variation of the Turing test, in which participants were asked to say just one word to prove their humanity.
What would you choose?
Developed in 1950 by British mathematician Alan Turing, the Turing test is a way to determine a machine’s capacity to exhibit intelligent behaviour equivalent to, or indistinguishable from, a human being.
In the test, a human evaluator assesses a text-only conversation between a human and a machine.
Based on their interaction, the evaluator would have to distinguish which was human, and which was artificial intelligence.
The AI only passes the test if the evaluator can’t tell the difference between the two.
(Sidebar: For a great movie on the topic, check out Ex Machina.)
Researchers John McCoy and Tomer Ullman from the Department of Brain and Cognitive Sciences at the Massachusetts Institute of Technology have created a slightly different version of the test.
Instead of holding a conversation, participants must use a single word to convince the evaluator of their humanity.
No pressure, huh?
The results of their test were published in their paper, “A Minimal Turing Test”.
From around 1,000 participants, the word “love” was the most popular, with 14 per cent using it.
It was followed by “compassion” (3.5 per cent) and the mundane “human” (3.2 per cent).
I mean, only a human would use “human” to prove that they were, indeed, human.
These answers sound kind of unimaginative, right?
Don’t worry.
Not everyone was a bore.
There were plenty of amusing picks like, “serendipity”, “moist”, “supercalifragilisticexpialidocious” and “d’oh!”.
The second part of the study involved determining which of those words would most effectively convince other humans that the respondent was not a robot.
Researchers proceeded to test the words on a new group of more than 2,000 participants.
The second group of volunteers were given pairs of words and told that one of them was chosen by a machine and the other by a human.
The trick was on them — all of them were suggested by the previous pool of human volunteers.
What the study found was that the word “robot” was the most recurrently flagged as coming from artificial intelligence.
Fancy that.
However, the best takeaway from the study is that the word “poop” was identified by the participants as the most likely to come out of a human.
“The high average relative strengths of the words ‘love’, ‘mercy’, and ‘compassion’ is consistent with the importance of the experience dimension when distinguishing the minds of robots and people,” write McCoy and Ullman in their paper.
“However, the taboo category word [‘poop’] has the highest average relative strength, referring to bodily function and evoking an amused emotional response.”
“This suggests that highly charged words, such as the colourful profanities appearing in Study 1, might be judged as given by a human over all words used in Study 2.”
The experiment provides some interesting insight into our human self-perception.
“Recall the word that you initially chose to prove that you are human,” they wrote.
“Perhaps it was a common choice, or perhaps it appeared but one other time, your thoughts in secret affinity with the machinations that produced words such as caterpillar, ethereal, or shenanigans.”
“You may have delighted that your word was judged highly human, or wondered how it would have fared.”
“Whatever your word, it rested on the ability to rapidly navigate a web of shared meanings, and to make nuanced predictions about how others would do the same.”
“As much as love and compassion, this is part of what it is to be human.”
It begs the question: what word would you choose?
* Benjamin Piñeros is a filmmaker and writer. He tweets at @IamPineros. His website is estamosobservando.com.
This article first appeared at www.techly.com.au.
