The field of Artificial Intelligence (AI) has seen tremendous growth in recent years, with advances in machine learning, computer vision, and natural language processing leading to significant improvements in the capabilities of AI systems (I mean, just look at the recent advances in generative AI…). However, while these advances have enabled machines to perform a wide range of tasks, they do not yet provide a complete understanding of how human intelligence works, not to speak about coming even close to it. This is where cognitive architectures come in.

Cognitive architectures are computational models that aim to represent and simulate the workings of the human mind. By creating these models, researchers can gain insights into how the mind processes information and make decisions, which can in turn be applied to the development of more advanced AI systems.

There are several popular cognitive architectures that have been developed over the years, each with its own strengths and weaknesses. For example, SOAR is a symbolic architecture that focuses on problem-solving and decision-making processes, while ACT-R is a more general architecture that aims to represent a wide range of cognitive processes.

One of the key benefits of cognitive architectures is that they provide a theoretical framework for understanding how the mind works, which can then be used to guide the development of AI systems. This is especially important when it comes to developing systems that are capable of human-like thinking and decision-making. Just think for example of chatGPT… Is it fooling you with giving the impression that it is smart? Not for long, I imagine. Indeed, there are some analogies that can be drawn between human brain areas and large language models (LLM): there is some kind of memory (hippocampus), and some attention (neocortex). But for LLMs, this rather corresponds to some kind of puree of the human brain – and you see this in its answers.

But by using cognitive architectures to study the inner workings of the mind, researchers can gain a deeper understanding of the underlying processes that are involved in human cognition, and use this knowledge to inform the design of AI systems.

Another important role of cognitive architectures is in the evaluation of AI systems. By creating models that accurately represent human cognition, researchers can use these models to assess the performance of AI systems and identify areas for improvement. This can be particularly valuable when it comes to developing systems that interact with humans, such as personal assistants or conversational agents.

Despite the many benefits of cognitive architectures, there are also several challenges associated with their development. For example, creating a complete and accurate representation of human cognition is a complex and difficult task, and requires a deep understanding of the workings of the mind. Additionally, the field is rapidly advancing, and new research is constantly shedding new light on the nature of human cognition, which can make it difficult to keep up with the latest developments.

To wrap it up, the role of cognitive architectures in research on strong AI is significant, providing researchers with a theoretical framework for understanding human cognition and guiding the development of potent AI systems. While there are challenges associated with their development, the benefits they provide make them an essential part of the AI research landscape.