I read this book in 2023 after the hype around chat-GPT inspired me to start reading and researching Artificial Intelligence. As I commented in my blog posts about my original exploration of chat-GPT, the technology isn’t quite there, but even as I wrote this article in June 2024, the large language models have evolved significantly.
The title of the book hints that it revolves around one of the big questions of the moment: Can Artificial Intelligence ever be sentient? The first step to answering this question is how you define sentience.
What is the definition of Sentience?
In his book The Sentience Machine, Amir Husain defines sentience as being self-aware, which gives one the ability to set meaningful goals.
While Joanna J Bryson defined sentience as having the perspective of feeling from an article titled One Day, AI will seem as human as Anyone. (Link to article)
But that on its own isn’t enough. How do you know that I’m sentient? How can we know if a piece of technology is sentient? The book doesn’t give an answer to this question, but with a definition of sentience, we might be able to consider what evidence you would look for.
Humans and Artificial Intelligence both run on a Neural Network
In 1960 Henry Kelley and Arthur Bryson developed a Learning Algorithm that would work in a Neural network. In 1989, George Cybenko, who was a researcher at the Centre of Supercomputer Research and Development, proved that a neural network with enough neurons and data “could approximate a mathematical function with arbitrary percussion.”
The Human brain is also a neuron network with approximately one hundred billion Neurons. On average, each of these neurons is connected to another 1,000 via a junction called Synapse. This network encodes most of the information held within a human brain. The human brain is highly parallelised, allowing it to process multiple pieces of data at the same time, though it does have a slower clock speed when compared to silicon-based processors
The human brain is very good at Cognitive tasks, but it is less good at “purely mathematical tasks” and abstract thinking.
If Artificial Intelligence is using a neural network like my own brain, and I am a sentient human being, then why can’t It become sentient like me? How will we even know?
Conclusion
I would recommend this book to anyone who is interested in Artificial Intelligence and its likely future impact on society.
The Amazon associate link to the book can be used to purchase the book from Amazon. We will receive a small commission, which will help us.