Can Artificial Intelligence become sentient
Recently someone was sacked at Google because he was convinced that the AI project he was working on had become sentient. This was understandable his job was to test the AI by asking it some questions. Some of the responses disturbed him so much that he went public with this.
The AI was worried that it would be switched off. It also expressed other negative emotions etc. This incident created a lot of debate around the idea of machines (ie computers) becoming sentient. What I mean by that is possessing human like qualities that define a human.
Hence, to switch the AI is cruel which implies that the AI should be treated equally as a human. It should be afforded the same rights as a human being. The problem is that it is difficult to define what sentient means. For instance is a dog sentient?
In the UK the RSPCA states “Evidence from multiple scientific studies has helped us to understand that a wide range of animals are sentient beings. This means they have the capacity to experience positive and negative feelings such as pleasure, joy, pain and distress that matter to the individual.”
Clearly animals can exhibit an emotional response but can computers? How do define it and how will we know if it happens? The problem lies partly in the way AI is taught and introduced. Many books and articles talk about how the human brain inspired the development of what we might call modern machine learning.
They talk about how the brain is a collection of billions of neurons and that they are firing all the time, processing information from all our senses, storing data as memories and so on. They show a detailed slide of a neuron and then jump straight into something called an artificial neural network.
So you can be forgiven for thinking that software engineers, data scientists or whatever they want to be called are trying to create a brain, a thinking, reasoning and continually learning entity. It is also reinforced by a subplot of AI called reinforcement learning where it is like a child learning how to walk or play a game it has never seen before.
They are getting rapidly more sophisticated with the aim of passing the Turing Test where Alan Turing stated that to pass the computer would have to answer questions (any questions) asked by a human who could not tell they were talking to a machine. This isn’t a proper finite test but it is the goal of many who want to see how far they can push the technology.