Modelling Artificial Intelligence After the Human Brain

Artificial Intelligence (AI) simulates the process of learning that humans undergo when observing their surroundings. The software allows machines not only to process and store knowledge, but also to solve problems and make predictions of the future based on prior observations. In addition, AI helps machines optimize their process of learning based on prior experiences with learning. However, the main difference between AI and the human brain is that the human brain can immediately recognize and distinguish different physical objects, while AI requires several images of various objects in order to train itself to look for the differences between these objects. Researchers Maximilian Riesenhuber, PhD, and Joshua Rule, PhD, sought to design an AI software that could recognize different objects by placing them into different categories, rather than focussing on specific, easily-identifiable features, like shape and color.

Neural networks process observed information, or inputs, through nodes within multiple hidden layers before generating a result or classification, known as the output. Image courtesy of InnoArchiTech.

This program was an artificial neural network that modelled the human brain’s ability to create a hierarchy of knowledge and concepts that are used to identify different objects. The software focuses on identifying objects through high-level classifications rather than low-level observations. High-level classifications allow the computer to relate different objects that are visually similar to each other. For example, this program could recognize the similarities and differences between a zebra and horse in order to distinguish the two animals. In addition, it could place a platypus, a duck, a beaver, and an otter into the same category, due to their visual similarity, before distinguishing the animals through the specific features that are unique to each animal. This way, the program will not have to compare the given animal to every single animal in its entire store of knowledge. 

This process makes the thinking process of the AI more similar to that of the human brain by increasing its efficiency. Nevertheless, the human brain is still much better at learning through forming generalizations based on observing the traits of a few examples. However, as AI advances to become more similar to the human brain, there is potential to use AI to model and study the human brain in the future.

Works Cited

Riesenhuber, Maximilian. (2020). How the mind sees the world. Nature Human Behaviour. 4. 1-2. http://doi.org/10.1038/s41562-020-00973-x.

Rule, Joshua & Riesenhuber, Maximilian. (2021). Leveraging Prior Concept Learning Improves Generalization From Few Examples in Computational Models of Human Object Recognition. Frontiers in Computational Neuroscience. 14. http://doi.org/10.3389/fncom.2020.586671.