Turing Test in AI

The Turing Test in AI (Artificial Intelligence) is a method that is used to test the intelligence of a machine through specific inquiries. <!--more--> It tries to test whether a machine can think like a human. The Turing test uses three subjects in its performance, that is, a human, a machine, and an interrogator.

This article will give you a clear understanding of the Turing test in AI. Readers should have a basic understanding of artificial intelligence. Click here to read more on the introduction to AI.

Table of contents #

• What is the Turing test?
• Performance of the Turing test
• Requirements for passing the Turing test in AI
• Features of Turing test
• Limitations of Turing test
• Modern-day Turing test
• Conclusion

What is the Turing test? #

A Turing Test, as the name suggests, was introduced by Alan Turing in the year 1950, in his article 'computing machinery'. This test is aimed at testing whether a machine can think.

The test is based on a party game called The Imitation Game with its modifications. It involves three players: a machine (computer), a human respondent, and an interrogator.

The interrogator does the inquiry while the human respondent and the computer provide feedback.

The interrogator is usually isolated from other players to identify which among them is a machine. The players are allocated some characters as unique identifiers in their different locations.

The interrogator already knows that one of the players is a machine. However, a conclusion is made based on inquiries and responses from the participants.

Given that one of the players is a machine and with the ability to convert text to speech in the form of bits 0 and 1, the conversation is done using a screen and keyboard.

The interrogator might ask the following questions:

Interrogator: Are you a man?

Player x (computer): NO.

Interrogator: Convert the following binary digit into bits (4294967296).

Player X: Pauses for a while and gives the wrong answer.

From the above example, if the interrogator fails to differentiate between the computer and the human, then player x (the computer) will have passed the test successfully.

Research conducted shows that to date no machine has passed the Turing test.

Performance of the Turing test #

When the Turing test is performed, the human player and the machine are hidden from view to ensure clarity and transparency.

The interrogator makes inquiries to both the machine and the human simultaneously. He attempts to identify which among the players is the machine or human based on the quality of the feedback.

The test considers the computer to be an intelligent agent if the interrogator is unable to distinguish between it and the human player in terms of their responses.

The interrogator then concludes that the machine has passed the test successfully and, therefore, demonstrates human intelligence. In other words, it can think.

Requirements for passing the Turing test in AI #

To pass the test, a machine must meet the following requirements:

1. Natural language processing (NLP): It can understand natural human languages.

2. Knowledge representation (KR): The ability of the machine to store and retrieve information provided before or during the interrogation.

3. Automated reasoning (AR): This requires that a machine can use the stored information to answer questions and make new conclusions.

4. Machine learning: This feature is needed to adapt to new circumstances, as well as to detect and extrapolate patterns.

5. Vision: Vision is needed for recognizing the examiner's actions, as well as different objects.

6. Motor control: The ability to act upon an object as required.

7. Other senses: These include audition, smell, and touch.

Features of Turing test #

Turing Test requires the following:

1. Participants: The main participants include:

• An interrogator.
• A person (human).
• Computer.

2. The site: This is where the test takes place. The participants are kept in separate places.

3. The test: These are the questions that the respondents must answer.

Limitations of Turing test #

• The test requires physical interaction which necessitates the need for perception and actuation.

• The test is not reproducible. It uses a different code or logic from the original.

• Turing test may not be appropriate for analyzing the level of machine intelligence since it is not fully guaranteed that any computer that can showcase intelligent behavior is intelligent.

• It cannot be performed in an open place since one of its requirements is that the players should be in separate rooms to ensure they are hidden from view.

• The test is limited to analyzing human-like intelligence.

Modern-day Turing test #

The modern-day Turing test is an update of Alan's Turing test for testing the intelligence of an AI-based machine. The test aims at creating artificial intelligent agents that can communicate using different modes.

It also aims at understanding how far Natural Language Processing (NLP) has progressed in the past decade, as well as in the modern world.

It does not test conversational AI for intelligence but rather focuses on the effectiveness of conversational AI.

Several organizations are working to develop ostensibly human-like conversational AIs. A significant social competitive challenge will be to develop a test that determines how human conversational AI is.

Conclusion #

Turing Test is important in demonstrating the intelligence of a machine and whether it can think like a human.

This is usually done through interrogation. The test requires participants such as humans, interrogators, and a computer. It also requires a site where the participants are kept in a safe place.

Although the Turing test is resourceful in the tech industry, its validity has not been accepted by everyone. Passing this test remains a challenge to many artificial intelligence developers.

No machine has ever passed the test to date. Nevertheless, the Turing test has paved way for more research and innovation.

---

Peer Review Contributions by: Onesmus Mbaabu