Turing Test and Human Judgement

The Turing test has been an essential and influential concept in the philosophy of artificial intelligence ever since Alan Turing introduced the test in 1950 to test whether a machine is intelligent.

The Turing test itself is simple, but the underlying reasoning process of Turing is profound.

Thinking vs. Acting

Can machines think? If computers can reason, how can we tell?

Are trees intelligent? Can they think as well? How can we tell?

How about stones?

Well, the answer is we cannot tell. By common perception, trees are not intelligent and can’t think. What if trees and stones actually can think, but in a frequency that humans can’t understand and on a timescale beyond human comprehension?

Turing approached this question differently because thinking is challenging to define and measure. Instead of asking whether a machine can think, measuring a machine’s behavior is more manageable and realistic. If a device acts and behaves intelligently, then it is intelligent.

Humanly vs. Rationally

Now that we have established the acting and behavior of machines as the measure of intelligence, a follow-up question naturally emerges: What is intelligence? What do we mean when we say a machine is intelligent?

Intelligence itself is a generic word without rigorous definition. According to Merriam-Webster, Intelligence has the following meanings:

1. the ability to learn or understand or to deal with new or trying situations
2. the ability to apply knowledge to manipulate one’s environment or to think abstractly as measured by objective criteria (such as tests)

We have a general sense of intelligence. But what do we mean when we say artificial intelligence?

A quick response would be just like a human. If a machine behaves humanly, then we say it is intelligent.

The same ambiguity arises with acting humanly as with thinking. How do we define behave humanly for billions of people living on earth? And individuals behave differently, subject to their cultural, social, and economic situations.

Given that acting humanly is challenging to define, it’s reasonable to replace it with acting rationally, which generally means making optimal decisions with given inputs, a well-defined and researched field in math.

Plane vs. Bird

Modern people are used to airplanes and birds.

An interesting question is ‘why do planes look differently from birds?’.

Why don’t we design planes by mimicking the appearance of birds?

The goal is to design a machine that can fly rather than a machine that looks like a bird? The appearance of birds is just superficiality, whereas aerodynamics is the underlying truth.

The End

There are some fundamental truths in the designing of the Turing test and the designing of airplanes:

1. Judge an object or a person by its behaviors instead of thoughts.
2. Be cautious with words. Avoid being naive.
3. Avoid superficiality and seek fundamental truth.