AI: The very nature of human learning
Learning is the process of acquiring new or modifying existing, knowledge, behaviors, skills, values, or preferences.
Human Intelligence is an evolving process over a period of time. The very nature of the learning process of a human as a newborn kid starts with understanding what is happening around. We could evidently experience the anxiety of learning in a kid in its eyes. They get to describe what they are seeing and happening, through which they get descriptive learning.
After a certain period of time, the kid starts to learn why is this all happening. They don’t fear of failure at this stage, because they have never experienced anything as a failure. They would do all sorts of analysis to understand the cause or nature of a condition or situation, by which they get diagnostic learning.
Gradually as the years go by they develop the intelligence to relate things happening around by comparing the conditions or situation and foresee a probable outcome. They don't stop there. They let things happen, validate their foreseen probability and refine it through such predictive learning.
Through such a descriptive, diagnostic and predictive learning process, as a grown teen, they start to build an attitude or a point of view to each situation. They tend to create a set of definitive actionable activities from their learning to make things happen, through a perspective learning.
Finally as a grown adult, through all the above learning, the actions on a particular condition or situation become logical and very intuitive, through cognitive learning.
Artificial Intelligence is a process to build and launch a product, to undergo all these steps of the learning process.
Giving birth to a child, in the case of humans, is precisely the same as launching a product built to become artificially intelligent.
These learning can be done in three ways. Unsupervised, Supervised and Reinforcement.
In Supervised learning, the algorithm is fed with the desired output, before feeding the raw data for analysis. The algorithm is taught through a training data set that guides the machine. With the output of the algorithm known, all that a system needs to do is to work out the steps or processes needed to reach from the input to the output. It predicts based on various features.
In Unsupervised learning, the algorithm is not aware of what should be output. It has the ability to interpret and find solutions to a limitless amount of data, through the input data and the binary logic mechanism present in all computer systems. It always looks for a pattern in the given data. This fits well in the case of clustering and segmenting of data.
In Reinforcement Learning, is the algorithm which takes suitable action to maximize reward in a particular situation. It refers to goal-oriented algorithms, which learn how to attain a complex objective (goal) or maximize along a particular dimension over repeated steps. It tries to figure on its own, and course correct on the way.
In essence, the training data is the textbook that will teach the AI algorithm to do its assigned task and will be used over and over again to fine-tune its predictions and improve its success rate. AI which has variables, algorithms that the data scientist has to adjust in several different ways, aimed at improving the accuracy of prediction.
The cleanliness, relevance, and quality of data have a direct impact on whether the AI program will achieve its goals. For e.g. give a student an outdated textbook and with half the pages missing, they won’t come close to passing the course.
Quite simply, the Artificial Intelligence product, after its initial launch, has to undergo several iterations of learning and tuning with adequate training data, so it behaves cognitively more accurate, to make incredible discoveries with high-order interconnections into a complex network. Towards the end of getting it 100% accurate, the real challenge lies in, how to build Artificial Intelligence with common sense and the ability to reason.
