Every time we use the term "machine learning", we find ourselves in the territory of neural networks and artificial intelligence. But that concept is hardly new. In fact, we have to go back to a time when the transistor was just beginning to leave the lab.
In this video, we see Claude Shannon, father of information theory, demonstrating the abilities of his little mouse Theseus, which can accumulate data, make mistakes, adapt to them, and ultimately apply the correct solution to escape from a maze.
The Technological Boom of the 1950s
The 1950s kicked off with a true explosion in technological development, despite the growing tensions of the Cold War. In 1947, Bell Laboratories built the first solid-state electronic transistor, and within less than four years it was in mass production. Some of the brightest minds in the world were working for Bell Labs at the time, and one of them was Claude Shannon.
His groundbreaking 1948 paper, "A Mathematical Theory of Communication", was revolutionary, introducing concepts such as redundancy and entropy. But he didn't rest on his laurels. Another of his creations was Theseus, which, thanks to the AT&T archive, we can see in action:
How Theseus Learns
As Shannon himself explains, Theseus is an electromechanical mouse that can "learn from past experiences", in other words, adopt a trial-and-error methodology to escape from a maze divided into 25 sections. Each time it finds the correct path, Theseus records the information in its 'memory' (technically, a system of telephone relays in the background).
If it returns to the start, all it has to do to reach the exit is replay that accumulated experience. But the best part is its potential for adaptation. When it detects changes in the maze walls, Theseus partially replaces its old data with new parameters and applies the rest of its knowledge without any problems.
As a true scientist, Shannon can't help but explain that the work is not done by Theseus, but by the system installed underneath the table, with an electromagnet, two motors, and switches to determine its location. The exploration method used by Theseus may not be the most efficient, but it guarantees finding the exit in any maze.
Source: AT&T Tech Channel
