How can a Machine learn? The Key to the future. A blog for DUMMIES!

How can a Machine learn? The Key to the future. A blog for DUMMIES!

Machine learning in a nutshell

The Evolution

In the past, chatbots were primarily designed to automate responses by offering users a selection of prompts to choose from, which would then generate a corresponding output. However, contemporary chatbots like ChatGPT have evolved far beyond this simplistic model. They not only allow users to ask questions in any language or style they prefer but also assist with tasks such as homework, code debugging, and even engaging in interactive activities. Whether you're an expert or a novice, have you ever pondered how chatbots transitioned from being confined to specific inputs and styles to becoming versatile tools capable of answering any question and aiding in tasks like crafting blog posts?

A side fact, the first Chatbot created was made by Joseph Weizenbaum in 1966 to be used in psychotherapy.

What is Machine learning?

Well, from the name, it is basically a machine that have the ability to learn by using certain Algorithms which enables the Computers to learn from data provided and training. There are mainly three types of Machine learning approaches, supervised learning, unsupervised learning and reinforcement learning.

Supervised Learning

In supervised learning, an input (which is called features) is given with it's output -(which is called label to the model. The model studies the label and the feature together. The goal is to learn a mapping from input to output so that the algorithm can make accurate predictions on new, unseen data. Common tasks in supervised learning include classification (predicting a category) and regression (predicting a continuous value). Didn't understand? let's look at an example. Let's say that a model is designed to identify whether the animal provided is a cat or a dog. The model is given set of pictures(inputs) and labels(dog) to study their features --ears, noes, eyes ,etc. Now after learning how does a dog or a cat look like, it will then be tested. New pictures of dogs and cats will be given, and after learning their features, the model can decide whether the picture provided is a picture of a dog or a cat. Supervised learning can be used to give predictions or in Chatbots like ChatGPT( It was trained on a vast amount of human response)

Unsupervised Learning

In unsupervised learning, the algorithm is trained on a dataset consisting of input data only, without any corresponding output labels. The goal is to discover patterns, structures, or relationships in the data without explicit guidance --unlike supervised learning that requires output data to be given. Let's say that you have a group of pencils and pens mixed up together, and you are told to organize them. Let's also say that you don't know the difference between a pencil and a pen. The only approach to solve that problem is to look at each one individually and group the ones that looks a like together, and study the difference by yourself. It can be used in text generating AI.

Reinforcement Learning

In reinforcement learning, the algorithm learns to make decisions or take actions in an environment to maximize cumulative rewards over time. The algorithm interacts with the environment, taking actions and receiving feedback (rewards or penalties) based on the consequences of its actions. The goal is to learn a policy that specifies the best action to take in a given state to achieve the highest cumulative reward. Reinforcement learning is commonly used in applications such as game playing, robotics, and autonomous systems. A Robot Vacuum, for example, uses Reinforcement learning in order to discover a room. And whenever it falls into an obstacle, it learns from that experience and receives a negative feedback.



Don't mix Artificial Intelligence and Machine Learning Together!

Artificial intelligence (AI) refers to the development of computer systems that can perform tasks that typically require human intelligence. However, it is not same as machine learning.

As mentioned earlier, Machine learning is mainly about learning and training on a large amount of data, without the need to be explicitly coded. Artificial intelligence is mainly about solving problems that requires human intelligence by using different approaches, it can be rule-based, heuristic-based, or use machine learning techniques to achieve their goals. Examples of AI systems include virtual assistants (like Siri or Alexa), autonomous vehicles, facial recognition systems, chatbots, and recommendation systems. That means, AI models may not use artificial intelligence.

How does the future look Like?

I Guess we are living in the Future! On November 30, 2022, Chat GPT was released. After its release, the AI technology has Exploded! New online and free image generators came to life, which can generate any idea written by the users in a matter of minutes. In addition to some new video Generators like DeepFake. 

That prompts further inquiry: Is the development of artificial intelligence spiraling beyond manageable bounds?
Back to blog

Leave a comment