As the years progress and new developments emerge, the progress of AI has reached the point of having the ability to process more information than humans, at least in terms of data volume. However, this responds to a kind of reasoning, there are also others that are of superior domain of the human mind, such as reasoning by analogy.
An investigation started at the University of Amsterdam, presented an analysis proposal that addresses this way of reasoning, from the perspectives of the human mind and artificial intelligence systems, in order to dimension those capacities and contribute to the development of both.
Analyzing Analog Reasoning
Reasoning by analogy, which allows you to think of solutions to new problems, based on the experience acquired with previously known situations, is one of the weakest points of artificial intelligence today.
Investigating the presence of this element in AI systems and in children, Claire Stevenson, assistant professor of psychological methods at the University of Amsterdam, also trained in mathematical modeling and computer science, presented a study that, in her words “Try to test human intelligence in AI, and test AI intelligence in children”, in order to generate mutual feedback.
After starting his career in the field of developmental psychology, where he investigated the learning potential of children through the development of analogical reasoning, Stevenson later delved into the idea of applying mathematical models to measure creative processes, applying his studies in information science.
“The focus of my research is now shifting to cognitive AI and the imitation of human intelligence. I am exploring algorithms and the extent to which they can resolve analogies, in other words that thirst is to drink as bleeding is to bandage. My colleagues and I are trying to answer the question of how much intelligence really is in Artificial Intelligence », he pointed the researcher in statements collected by the portal of her university.
While the strength of AI focuses on the knowledge acquired and procedures learned, such as arithmetic operations, the skills associated with the so-called “fluid intelligence”, which include reasoning and problem-solving skills, continue to be the predominant domain of the human mind,
‘AI machines and algorithms have enormous storage capacity, much larger than human memory, and can retrieve and process information at lightning speed. They can do some amazing things », highlights Stevenson, “But this first form of intelligence is actually quite simple compared to the other, which the AI is still struggling with”he added by way of contrast.
Stevenson’s research plans to compare the way in which more complex reasoning tasks and exercises, such as bongard problems.
After the probing stage, the researcher hopes to apply this knowledge for the subsequent development of artificial intelligence systems and learning environments for children.
Imagine what would happen if AI mastered analog reasoning and learned to think more flexibly and creatively. You could combine that ability with your higher general (factual) knowledge and processing skills to identify relationships between highly diverse and seemingly unrelated subjects. For example, AI could identify parallels between the course of a disease and its recovery and the fight against climate change, and provide unexpected insights to help us solve complex problems “commented the teacher.