Is AI Intelligent Part 1?

We live in a period of peak development of "artificial intelligence," the highest in the history of mankind. AI is currently experiencing a classic "hype phase," as described by Gartner. The changes in education and information perception brought about by AI can be compared to the Industrial Revolution.

Artificial intelligence as a concept goes back a long way. The AI development graph resembles a wavy curve, with periods of growth and decay. Bursts of development activity are usually associated with either new algorithms or improvements in computing technology. However, after each such surge, a period of stagnation begins, called the "winter of artificial intelligence." The intervals between these periods of activity and stagnation are shortening, indicating an acceleration in the pace of development in this area.

When the limit comes, and technology seems to hit the ceiling, the mental activity of humanity begins to work with renewed vigor: "How can we overcome these limitations?" Humanity has shown its ability to overcome anything, so any such "winter" about AI has previously been overcome.

The next significant peak on this path should be the creation of AGI - general, or so-called robust and artificial intelligence. However, today, many scientists, guided by their understanding of algorithms and mathematical theorems, express doubts that even the most powerful computing technology will achieve accurate AI self-awareness comparable to that of humans.

The critical question: computing resources + storage capacity = Intelligence?

Progress in artificial intelligence is associated with translating phenomena from the social human sphere into mathematical language. The formalization of, for example, human relationships, social interactions, and linguistic processes gives rise to the innovations we see today. However, the critical role is not played by mathematics itself but by the ability to translate different areas of knowledge into mathematical language.

The approach to defining general AI and its capabilities is instead associated with philosophical and religious categories, such as the concept of human consciousness and free will. Today, many developers of artificial intelligence technologies have a fundamental education only in mathematics, which raises questions about the correct direction of their research in general or strong AI.

Is it possible to achieve AGI using math alone?

Comparing the study of the brain with observing the operation of a computer, one can conclude that trying to analyze the electrical activity of neurons is like trying to understand the logic of a computer by observing its electrical signals. By studying only the physical processes in the brain, it is tough to understand the reasoning of thinking or the essence of consciousness.

Although modern technologies allow us to analyze and reproduce speech patterns, turning them into voice assistants, the soul and feelings still need to be translated into mathematical models.

Thus, understanding the functioning of the brain requires not only a physiological but also a philosophical approach.

Yes, we have learned to emulate (to an extent) emotions in machines. But what emotions are and how to digitize them is still unclear. Their nature is apparent to psychology, philosophy, religious studies, and biology specialists, but they need to learn to combine all these interpretations in one theory and create a digital analog. Without it, it is impossible to fully understand human intelligence, which means it is impossible to create true artificial intelligence, that same AGI. Using "fuzzy" logic can be a promising first step in this direction.

The real breakthrough in creating AGI will come when systems begin to sense and go beyond the simple Internet of Things. The system's ability to store human perceptual experiences, including emotions and memories, may be key. However, there has yet to be an AI model that has demonstrated this capability.

If systems learn to capture, store, and process our psychological memories constantly, they can become fundamentally more advanced than current models based on fixed data sets.

Finally, if we do not have AI, then what do we have?

A - Automating new, more complex processes (e.g., image classification, textual information extraction, facial recognition, etc.)

S - substantially higher speed of existing processes

M - significantly more effective information management (amount and types of information we can manage)