Posted by When I was a kid, I really wanted to build a robot that would drive around the house fetch objects, grab a ball or respond to spoken words. All these abilities were absolutely beyond the reach of home electronics and mechanics. I built electronics from a schematic from the Young Technician magazine, and mechanics from LEGO bricks. At that time there was no Internet or cell phones, not to mention Google or Apple assistants.
Several decades have passed, and today we can confidently say that an artificial manager would not only do all the things that were a dream back then, but we will also equip it with artificial (?) intelligence. So at what levels can we consider building such an artificial manager?
Level 1.
We can build a manager in the form of a knowledge management system, in which methods of inference in a given situation are built in. These are fairly simple systems, similar in complexity to a washing machine or an electric kettle. In this regard, we use ordinary pattern recognition, known until more than 40 years ago. What happens to the team when we present them with an artificial manager built about such principles, I will present in subsequent posts. I have already built such a robot in 2021!
Level 2.
The next level of sophistication is the operation of Google or Apple assistants. These programs handle context cognition and draw simple conclusions from situations. Similarly, an autonomous vacuum cleaner or cars with automatic parking function work. Technically, we build such devices about machine learning methods.
Level 3.
If we apply deep learning, we have programs for analyzing images and recognizing diseases (IBM Watson) or playing the game GO. We are then talking about artificial intelligence in the narrow sense. In this way, our artificial manager would already be a pretty tolerable manager. But how about going a level higher?
Level 4.
Let’s teach the robot to infer the artificial manager from the context, and let it make decisions based on the knowledge it gathers about the world, not just the knowledge it gathers from the activities of human managers. This is a much more difficult task, and we will be working on it in the coming months. I posed a research question: what would happen if ChatGPT knew everything about how a single team manager behaves in a given project – knew at least some of his activities? Would he be able to manage another similar project just like the original man?
What results this experiment will bring, I will write in future posts.
To see more about the levels of artificial intelligence, you can watch a video on YT: The 10 Stages of Artificial Intelligence