On Cybernetics and its influence on AI

In the first blog post of the “History of AI” series, we covered the founding event of Artificial Intelligence. In 1956, Artificial Intelligence was born as a discipline. Many events and discoveries have led to this, one of which is undoubtedly the publication of Wiener’s Book: “Cybernetics: Or Control and Communication in the Animal and the machine” in 1945. 

This blog post will give a short introduction into the topic of cybernetics, where it comes from, what it means and of course, how it relates to AI.

‘Norbert Wiener’s now-classic work entitled The Human Use of Human Beings: Cybernetics and Society reveals the thesis that “society can only be understood through a study of the messages and the communication facilities which belong to it”

Broadhurst, A. R., & Darnell, D. K. (1965). An introduction to cybernetics and information theory.

What is Cybernetics?

The term cybernetics was coined by Wiener’s publication, but the words origins are Greek and the idea it describes is also not fundamentally new. So, let us start at the beginning: the Greeks. The term “cybernetics” comes from the Greek word “kubernetes”, which translates to “steersman”. It is the same word root as in the word “govern”. Plato already talked about the concept of government in terms of control. In the end, a government is steering the country.

The term cybernetics comes from the Greek word kubernetes, to steer. We are constantly steering ourselves through our environment, much like a helmsman steers their ship.

Cybernetics is a school of thought about control (hence the relation to “steering”) and communication; a philosophy, a language that describes systems that have a goal. Notice the vague formulation of “system” here. The crucial part of this theory is that it does not differentiate between human or animal. Not even between animal and machine. Cybernetics sees everything that pursues a goal to be essentially the same. The components may differ, but the mechanisms are the same.

Action, Sensing, Comparing

But what are these mechanisms which appear in both human, animal and machine? It is a loop of action, sensing and comparing which every goal-driven system follows. Any action we do can be broken down into these three fundamental steps. Take for example the action of putting bread in the toaster. We grab the toast and move it over the toaster (action). With our eyes, we sense if the current position of our hand allows us to drop the toast in the toaster (sensing). If yes, we drop it into the toaster and press down the switch. If not, we correct our hand position and drop it. We compared our sensory input (what we see) to our goal (putting the bread into the toaster so we can finally have breakfast).

Now comes the interesting part. Once you press down that toast, the toaster starts acting by browning the toast. Through metal wires and current flow, it senses how much time has passed yet and releases the toast if the perfect-toast-time is reached or continues to brown it if not. Once again: different components, same mechanism. 

Communication is the key

The main point of cybernetics is that all of this is only possible through communication. Without communication, no goal could be reached. We all have a basic grasp of what happens in our body that allows us to put the toast into the toaster. 

But also inside the toaster, communication happens to achieve the goal of a perfectly browned toast, just through current and metal wires instead of neurons and muscles.

If you see the world through a cybernetic lens, you will notice that basically everything can be seen in a cybernetic way. It is not only the living world, humans and animals. Beyond that, cybernetics can also be applied to social systems. During a conversation, we act by speaking. We sense the answer of our conversation partner, verbal and non-verbal, and compare if the conversation is going in the direction we want it to. If not, we try to steer it in the desired direction through our next action.

In our blog post about the Turing test, we discussed already what controversial topic the combination of machinery and intelligence was. And Turing’s paper was published five years after Wiener’s book. His claim that humans and machinery are not fundamentally different, was revolutionary, if not scary for people at that time. But it was also part of a shift, which brought the whole field of Artificial Intelligence alive. 

From Cybernetics to AI

Once the idea that humans and machines are essentially the same became more popular in the researching community, scientists began to seriously consider what it would need to develop machines with human-like intelligence. 

And the groundwork for this was done already. Prior to Wiener’s publication, Turing introduced the Turing machine, also known as the universal computing machine (1937, read more in our post about him here), one of the most influential works in computer science and mathematics. And that before the first digital computer was built. In 1943, McCulloch and Pitts proposed an artificial neuron, which by some is considered to be the first work in artificial intelligence (we briefly touched on this in the most recent history of AI blog post, read it here). It was the first theory of mind and brain in mathematical notions. Once it was agreed that both systems have the same expressive power, the barriers between machine and mind were broken.  

From there onwards, the interdisciplinary exchange continued, mind and machine were brought together closer and closer. The focus of cybernetics on communication was addressed in Information Theory by Claude Shannon already in 1948. Paired with other advances in the field of computer science, the path was paved for the Dartmouth conference, AI became its own discipline. Exploring ways to achieve human-like intelligence with machines was now its own field of research. 

But also the other direction was explored. Scientists started to see computers as a way to understand the brain. Not only did Wiener’s theory reduce the barrier between human and machine, but also helped to demystify the basic mechanisms of our brain and all purpose-driven systems in general. Cybernetics and control theory finds, to this day, wide application in various disciplines, including environmental, social, learning and management. 

Wiener on the future

Wiener expected that machine to machine and human to machine messages would be steadily increasing in our society. According to him, it does not matter for my own goal achievement whether I give my instructions to a machine and await its output or a human. All that matters is communication. Inside me, between me and my potentially mechanical communication partner and inside them. The basic principle of cybernetics is simple, yet effective and its implications have been immense. It was part of a major shift of thinking which contributed to the digital revolution. And in my eyes, time proved him right. 

References