top of page

The Unexpected Learning Gap Between Children and A.I.

Frankie Reyna

With recent advancements in artificial intelligence appearing eerily human, we are left to wonder: what really differentiates us? While artificial intelligence holds a clear advantage in many aspects of our lives, there is one intriguing area where we differ: our inherent ability to learn connections and think abstractly. This distinction is fascinating due to its stark contrast with artificial intelligence learning, illuminating our incredible ability to adapt, learn, and extrapolate cues from our environment. In this article, we will explore the differences between human learning and artificial intelligence learning, and delve into the disparities in our thinking and reasoning. But first, let's examine what makes humans special.

          Our brains are incredibly adaptive. Every day, we form new memories, gain new knowledge, and use that information to make inferences about the world around us. This adaptability is critically important in the field of artificial intelligence, where attempts are often made to recreate human neuroplasticity. Yet, our current neural networks often need to be pre-trained on existing data. Children, on the other hand, can assign new labels and derive abstract meaning for entirely new objects without prior knowledge or training. This capability suggests there's something we possess in our perception of the world that AI can't replicate, specifically the ability to form abstractions from patterns.  

         Stuart Russell, a professor of neurological surgery and computer science at Berkeley, highlights this problem. Russell explains that AI 'cannot build complex knowledge structures from text; nor can they answer questions that require extensive chains of reasoning with information from multiple sources.' While artificial intelligence may be able to develop pattern recognition with extensive training, it often struggles to deeply understand and interrogate information from the environment in order to make new inferences. Artificial intelligence does not possess a true understanding or the ability to infer relationships beyond what it has learned from data.


Artificial intelligence does not possess a true understanding or the ability to infer relationships beyond what it has learned from data.

What makes our human brains so different from artificial intelligence, and from other lifeforms for that matter? The answer, like many other things, comes from our hundreds of thousands of years of evolution. Our "developmental brain" is a significant contributor to our capacity for abstract thought. fMRI studies reveal that these areas of the brain are active during "task-negative times," when the subject is engaged in self-referential processing. This process creates connections and facilitates pattern learning, capabilities found to be far less developed in other animals and primates. Despite its significance, we still have much to learn about this unique part of the brain that handles our abstract thinking and connection learning. What implications does this hold for artificial intelligence? When it comes to our current knowledge, we can infer little from the brain's workings. The mechanisms of the developmental brain are not yet fully understood, making it immensely challenging to replicate and implement such concepts within artificial intelligence. Despite this, artificial intelligence has made significant strides, particularly with the development of technologies like LSTMs. These advancements provide AI with an additional advantage and spark both hope and fear about the potential of one day creating an AI capable of abstract thinking.    



Grötker, Ralf. “Abstract Intelligence – How to Put Human Values into AI - Netopia Netopia.” Netopia, 27 March 2020,


Raichle, Marcus E. “From the Cover: The human brain is intrinsically organized into dynamic, anticorrelated functional networks.” NCBI, 23 June 2005,

Serene Gradient
bottom of page