Pretext: “Intelligence is the ability to acquire and apply knowledge”. Machines are made intelligent when they are trained to learn patterns using human intelligence through Artificial Intelligence. What if a Machine can train the other Machine?
We have heard about Robots assisting humans in most tasks. Wherein a software engineer trains a machine learning algorithm about patterns and methods so that the robot can perform the task again. Robots store this learned behaviour in a central repository called RoboBrain that’s accessible by other robots. However, with Robots came many technical challenges which are still in the process of transformation.
Today, talking about machine learning is a widespread phenomenon. A significant example of machine learning, which received highlight in 2015, was a driverless car. It is a car which understands the driving patterns of its owner and gets trained over a period. Tesla launched Model S with the vision to provide driverless car assistance to its owners. Surprisingly, after some time the car itself learnt the driving skill and route options. “Each car could improve its own autonomous features by learning from its driver, but more significantly, when one Tesla learnt from its own driver—that knowledge could then be shared with every other Tesla vehicle”. – Tesla CEO Elon Musk.
Navigating through technological transformations, one can determine that the speed at which machine learning software is working, it is going to create maintenance challenges. The systems are improving exponentially.
What comes as more surprising, is the concept of Machines communicating gained knowledge and teaching other Machines. Regardless of the progress, the competition between two machines is the darker side of the development.
The training of data is not an easy deal for machines. The data is the raw material for this practice. However, the data available is unclassified. The variance in the data creates asymmetrical experiences. When the object in question is a machine, the variable data can create an array of misinterpreted information and loss of previously stored information. This situation will impart a significant effect or defect in some cases.
A machine learning knowledge from a central repository located miles away, and performing tasks using the same knowledge, is incredible.
For example, one driverless car may take significant time to learn to navigate a particular city, while one hundred driverless cars can navigate that same city together. Using and sharing the knowledge they learnt, can drastically improve their algorithms in terms of quality and time.
The future of Artificial Intelligence resonates the useful information sharing between machines following common patterns and goals. It will be interesting to see if someday, the same machine algorithm can be used for performing different tasks with different goals. For now, all AI-powered devices continue to leverage this shared knowledge transfer. This enables faster learning, taking the pace of development to another level.
Well, we must consider that today we have only started with these technological advancements, as most of the projects are running in their test phase. We are yet to witness the wonders of technology infused with human skills.