Cognitive AI: The Road To AI That Thinks Like a Human Being

One of the most persistent myths surrounding the advent of Artificial Intelligence lies in the fantasy that advanced new technologies carry the "superpower" of being able to address every subject. For some, AI is providential, while for others, it is nothing less than a threat to humanity.

In reality, artificial intelligence technologies cannot solve problems for which they were not designed.

Narrow AI vs. General AI

At present, we only really have full access to "Artificial Narrow Intelligence (ANI)," which means it only focuses on a single problem.

The fact remains that ANI systems are capable of performing tasks much faster than any human being, which is a great help in terms of our overall productivity, efficiency, and even quality of life. ANI systems like IBM's Watson, for example, are able to harness the power of AI in medicine to help make data-driven decisions, making care faster, more accurate, and safer.

Experts around the globe are striving to develop Artificial General Intelligence (AGI). This ultimate form of AI aims to build intelligent systems capable of handling any task or problem in any field. To date, there are no truly general AI systems.

AGI remains in the realm of research (and science fiction!). In fact, the theoretical performance of these systems would be identical to that of a human being. Thus, because of their ability to access and process huge data sets at incredible speeds, they would be in a position to surpass the overall capabilities of humans.

Today, AI performs tasks such as visual perception, voice recognition, optimization of decision-making processes, text generation, and translation between languages quite well.

"Intelligence can be defined as the ability of an organism to represent information symbolically, using a process of categorization, in order to understand the context in which it acts, to be able to reason, to make decisions or to solve problems" — Mark Pohlmann — CEO AETOS

Cognitive AI

To go a step further, research in the field of cognitive AI integrates different modules for perceiving the world, which is the basis for acquiring knowledge and, therefore, for learning. This work focuses on the use of models to simulate the human thought process in complex and changing situations where responses can be ambiguous and uncertain.

But it's important to keep in mind that today, the quality of an AI program relies on the data on which it is developed and trained. Humans have yet to define the cases and scenarios in which the AI program will work, however complex. An AI program will work within these cases and scenarios, but it doesn't define new ones.

Unless...he was able to adapt.

Adaptive AI

According to Gartner, "Adaptive AI systems support a decision-making framework centered around making faster decisions while remaining flexible to adjust as issues arise. These systems aim to continuously learn based on new data at runtime to adapt more quickly to changes in real-world circumstances."

The aim of adaptive AI is, therefore, to bring these capabilities to a machine, enabling it to adapt to the environment in which it operates in order to remain flexible and able to reason, make decisions, or solve problems.

Mark Pohlmann explains that, to make adaptability possible, "it is essential to provide the machine with a data structure analogous to human long-term memory. The information must be represented symbolically and categorized, and manipulated objects must be linked to one another in such a way as to enable inheritance of descriptive, functional, dynamic, and symbolic properties."