On Intelligence | Jeff Hawkins

Summary of: On Intelligence
By: Jeff Hawkins

Introduction

Embark on an insightful journey as you explore the realms of intelligence in the human brain, and how it contrasts with the artificial intelligence of powerful computers. In the book ‘On Intelligence’, author Jeff Hawkins takes us through the fascinating complexities of the human brain’s neocortex, responsible for our sensory perceptions and conscious thoughts. Delve into the topic of neural networks, and the attempt to create machines that emulate the human brain, while identifying existing limitations and potential improvements. This summary will reveal the astounding potential of intelligent machines and their possible positive consequences for humanity.

Computers vs. Human Intelligence

Computers can never match human intelligence, regardless of their processing power, because they are limited to pre-programmed tasks while the human brain can learn and understand new concepts.

As computers have become more sophisticated, researchers have envisioned the possibility of creating a computer that could think like a human being. This concept, however, is not plausible since the fundamental principles of computers and the human brain are completely different. Although modern computers possess greater processing power than the human brain, they are still far from being truly intelligent. Unlike the human brain, computers are programmed for specific tasks, and that is the extent of their abilities. They do not have the ability to learn new things or incorporate new information to solve problems better in the future.

For instance, the renowned chess computer, Deep Blue, beat the world’s best chess player, Garry Kasparov. However, Deep Blue’s victory was not due to its superior intelligence. Kasparov understood the game’s inherent strategy, evaluated each move accordingly, and anticipated counter-moves instantly, which is not a capability of Deep Blue and most computers. A computer can only calculate probabilities and analyze every potential move, which does not imply that it “understands” the game or the world at large.

In summary, making computers more powerful or increasing their memory size is not equivalent to creating intelligent machines. Instead, it merely enhances their speed when performing pre-programmed tasks, which computers already execute better than humans. The initial step to developing a genuinely smart computer is understanding how the human brain works.

The Neocortex: The Gateway to Perception

The neocortex, a complex part of our brain, is responsible for sensory perception and conscious thought, and combines incoming sensory information with previous memories. The neocortex is made up of multiple layers, each adding increasingly detailed prior knowledge to the raw sensory information. When we perceive something, such as a familiar face, the neocortex connects the visual information with our memory of what human faces look like, allowing us to recognize the person. Our neocortex is so efficient that we are not consciously aware of this process, allowing us to experience the world seamlessly. If we encounter something new, the information goes to the top layer, storing the experience as a new memory. Our brains have an ever-growing database of things to compare new experiences to, thanks to the neocortex.

How Our Brain Predicts the Future

Our brain predicts the future by using interconnected memories to recognize patterns. When we experience something, different regions of the brain become active in a sequence like a pattern. When the same pattern is recognized, our brain knows what will happen next. Every new experience helps us adapt our predictions, making the brain a learning machine. This understanding can be used to create intelligent computers.

Neural Networks and the Future of AI

Scientists are exploring neural networks as an alternative to traditional computers in the pursuit of creating intelligent machines. The architecture of neural networks seeks to emulate the vast network of neurons that store memory and knowledge throughout the human brain. These networks operate by sending signals selectively via pulses of waves through interconnected neurons. However, artificial neural networks are still too simple to mimic the sophisticated functioning of the human brain since they lack the capacity for feedback loops and building memory banks. Yet, researchers continue to improve these networks to build truly intelligent machines.

Want to read the full book summary?

Leave a Reply

Your email address will not be published. Required fields are marked *

Fill out this field
Fill out this field
Please enter a valid email address.
You need to agree with the terms to proceed