Unraveling AI
Most people think that Artificial Intelligence is a new concept because it’s a term that, all of a sudden, can be found everywhere in almost every field. However, it’s actually been around for more than 60 years! Initially it was a discipline strictly associated with the technology world, but ever since the rapid advancement of technology, artificial intelligence is attempting to seep into everything that we do, which is why it’s become quite important to have a basic understanding of what it is and how it is going to change the world (hopefully for the better).
Along with Artificial Intelligence, another term that’s become popular is Machine Learning. Often the two are confused with each other, despite not being the same thing. Let me clear that confusion up by explaining what exactly Artificial Intelligence and Machine Learning is.
Artificial Intelligence and Machine Learning both encompass different technologies that one can spend years and years mastering. This article will just provide a brief introduction on each one. There will be future content that delves deeper into these intriguing topics, so don’t worry :)
What is AI?
Artificial Intelligence, in simple terms, is the intelligence of humans (natural intelligence) portrayed by machines — hence the term “artificial”. The word intelligence in this case refers to everything that a human can do — plan, understand human language, recognize objects, words and images, learn, make decisions and solve problems.
Any piece of software or device that possesses one or more of the aforementioned skills is said to be artificially intelligent.
There are two broads types of Artificial Intelligence –
1. Applied / Narrow AI (also known as weak AI)
As the name implies, anything that has narrow artificial intelligence has knowledge and expertise specific to only one task or area and lacks knowledge with respect to every other field.
For example, facial recognition software that can differentiate between a cat’s face and a human face. This software will know the exact differences between the features of a cat and a human and thus will be able to label a new image of a cat or human face accurately, but will not be able to predict the weather or mark emails as spam and not spam.
2. General AI (also known as strong AI)
This is designed to serve a larger goal. As opposed to narrow AI, generalized artificial intelligence is the representation of all human capabilities and more in a single device, that can, in the future, replace humans in a variety of intellectual situations.
So basically, a general artificially intelligent system would possess all the knowledge and capabilities that we possess, along with the advantage of processing and understanding this data much more efficiently and quickly than us. Such a system hasn’t been created yet, but there are hopes that this will become reality in the next few decades.
The future of generalized AI is all predictions at the moment, with the main hypothesis being technological singularity — the creation of super-intelligent systems that enable humans to transcend biological boundaries.
The best place to understand more about technological singularity is Ray Kurzweil’s book entitled “the singularity is near”. Click here to read more about it: http://singularity.com/
Click here to read more: https://hackernoon.com/general-vs-narrow-ai, https://blog.statsbot.co/3-types-of-artificial-intelligence-4fb7df20fdd8
What is ML?
Machine Learning is an approach to / an application of artificial intelligence. rather than being a separate concept of its own, as it is often considered to be, it is simply a way to achieve artificial intelligence.
Machine Learning is the process of a machine learning a particular task without being explicitly told to do so. Instead, it learns from past data to achieve the task. Hence, the primary principle of machine learning is for a machine to learn something without human intervention.
Take the example of the narrow AI facial recognition software that can differentiate between a human face and a cat’s face. How would this be made? Basically, the process involves feeding a program with millions of images labelled as “cat” or “human”. By going through these images and learning what features point to the label of “cat” and what features point to the label of “human”, the software is able to form an understanding of how the two are different. after this learning process is over, if the software is given an unlabeled image of a face, it can go through the features and correctly assign the label of either “cat” or “human — and there you have an artificially intelligent software formed by machine learning.
This is a very basic example, but this is primarily what machine learning does. there are many algorithms that perform the same task in different ways, and each algorithm is best suited for different situations. We will walk through these different machine learning algorithms in later articles, but for now it is important to know that they fall under three broad types –
1. Supervised Learning
2. Unsupervised Learning
3. Reinforcement Learning
Click here to read more: https://www.edureka.co/blog/what-is-machine-learning/
The topic of deep learning will be covered in future articles once the working of machine learning techniques is clear, but for now just know that deep learning is a subset of machine learning.
In a nutshell — Artificial Intelligence is the destination and Machine Learning and Deep Learning are paths to get there.
The next article will cover the different types of Machine Learning algorithms out there. Please let me know what you think of this article and how I can improve in the comments below! Thank you :)
