Turing Test - just as human ?

The question of what is intelligence has been subjected to deep research and analysis, both scientifically and philosophically, throughout history. From ancient Greece to the present, many thinkers have researched the source of this "natural intelligence" and have written about it.

Alan Turing was a valuable scientist who contributed to these "intelligence" studies and questioned this concept scientifically, and we continue to use the foundations he built in many technological developments today.

The main question was: Can a machine reach levels that can mimic human behavior? Is this possible? If possible, how should it be?

After a number of approaches, Alan defined “intelligent behavior” as the ability to acquire human-level intelligence during active conversation. The scenario was built on the interrogator's understanding whether the incoming responses came from a human or a machine. The interrogator had to question without seeing what was in front of him/her. The goal was to create a machine that could fool the interrogator. The text interface had to be chosen for this, because the presence of physical contact could cause errors and deviations in the experimental setup. Physical contact would change the original purpose. After all, the interrogator should not have obtained any information about what the other person/person-like is. In the experiment, the questioner had to communicate with two interlocutors, but the interrogator had to guess what/who the respondent was. If, during the test, the interrogator could not understand whether the answer belonged to the machine or the human, the machine that won the test should have been accepted. Although it may seem like an advantage given to the machine, it was actually the main mechanism of this test. If a “real person” can't understand that the answer is from a “real person”, then of course the machine is one step ahead. This is the fairness of the test.

Let's examine this scenario from the machine side. For a machine that exists only with a “binary system”, with a structure that works with “yes/no electricity”, it must have been quite difficult to maintain such a live conversation without being caught.

Before we explain what a machine needs, we need to examine how information or a thought pattern is formed at the thought levels of the human brain. It's impossible to map out the whole of it, but at least we have clear blueprints that science has given us. We see shapes - “geometrically”, we observe their ability to move - “kinematically”, we come into contact with and experience these objects - “physically”, we react or get a reaction to them - “behaviorally”, and finally we get a “cognitive” expression and we engrave this information into our brain as needed.

Let's revisit this sequence of operations like an algorithm and sort them:

• Geometric
• Kinematic
• Physical
• Behavioral
• Cognitive

As you can see, we can include all the interactions that most of us do in daily life. Within computer science we call it Cognitive Modeling, there is a field of study called Cognitive Modeling that deals with simulating the human thinking process. Studies conducted in this context aim to bring computer algorithms closer to human cognitive levels and organize experiments. It processes and tries to understand a person's approach to problem solving algorithmically. It creates simulations of human behavior with such results. Most of today's AI and DL algorithms are based on the products of this field of study.

Now that we have deduced the general schematic of the human cognitive approach, we can go back to the innocent machine in the Turing Test. What does the machine in this test need to achieve the human-intellectual structure we mentioned above?

• The machine must first contact the interrogator, understand the incoming inputs. It should deduce the general meaning of the incoming questions: Natural Language Processing
• The machine must store the information it obtains from this search for meaning. It should constantly compare with new incoming questions and update this database according to scenarios: Knowledge Representation
• As this stored internal information accumulates, the machine should derive a meaningful result from this accumulated data. Depending on the incoming reactions, this process of achieving meaningful results must be developed and maintained. A cause-effect mechanism must be created: Reasoning
• At the last stage, it should adapt to the general atmosphere and create and adapt action plans for new conditions together with all these fluid-living data: Machine Learning

With the above basic approaches, a machine is considered to be close to the intellectual-processing ability of a human, although it cannot fully reach it, it can reach sufficient capabilities for such testing.

This is the basic principle of the Turing test, and the aim is to test how much the gap between the human-machine understanding mechanism can be bridged. In the following years, the Turing Test was developed further and new combinations were added. In addition, this principle has inspired many game worlds and simulation theories. An interesting discussion is that this study causes many theological concepts to be questioned as well.

Personal Pages:

Go for -> Substack articles

Go for -> Donation page for studies