Things You Don't Know About Artificial Intelligence!

Humans learn from experiences and machines follow the instructions given by humans. But what if humans can train the machine and ask it to learn from experiences and act much faster for the next activity. It is Machine learning. Learning from experiences is an important thing. But it still requires reasoning and understanding. So Machine learning is a part of artificial intelligence. A computer algorithm improves itself from past data and experience. A computer discovers how to perform a particular task with the help of ML without detailed programming.

Most of us have come across online recommendations by Amazon, Netflix, YouTube, etc. It is possible by machine learning. Even ML assists in speech recognition, fraud detection, and in a self-driving car. Various sectors like Financial services, health care, oil and gas services, retail and transportation, etc. are using it on a wide scale.

The top five programming languages used in machine learning are Python, R programming language, Java and JavaScript, Julia, and LISP.

Aasim Khan and Siddharth Jain, students of class 11 of Jamnabai Narsee International School, Mumbai have developed a mobile-based application called RIDGE (Remote Identification and Detection of Genital Skin Cancer). They got Grand Awards at the US-based International Science and Engineering Fair (ISEF) in the Bio-Medical category. They were once in the Top 10 of the Microsoft Innovation Challenge. It is one of the largest global innovation competitions for students.

 

Understanding it through a Case study

Henry likes to listen to songs with high intensity and high pitch. A song is composed with high intensity and medium pitch. Now the composer wants to predict, whether Henry will like it or not. He will input the data of Henry's likes and dislikes in the machine. The machine will check past data and predicts whether Henry will like it or not. Hence, a machine can guide a person to make a product according to the customer's choice. More amount of data provides high accuracy in selecting the target. 

Machine learning is categorized as Supervised learning and Unsupervised learning.


Supervised learning

As the name indicates, a teacher or a supervisor will train the machine using well-labeled data. That means it is fed with the answer to some questions.

For instance: For training the machine with fruits, data will be like: 

If the object is red, spherical with depression at the top, then label it- Apple. 

If the object is yellow-green, long cylindrical shape, then label it- Banana. 


Credit: Ojas


Now the machine has learned from the data. When the machine sees the object, it will try to classify the object from the fed data. It will confirm the fruit as Apple (if it sees the apple), and put it in that category. In this way, the machine learns from the training data (data of fruit) and tests its knowledge on new fruit. From experience, the machine optimizes its performance. Accuracy in this type of learning is very high.

Application: Bioinformatics, speech recognition, spam detection, etc.

 

Unsupervised learning

Here, the machine is trained without labeled data. The algorithm works itself without any guidance from the supervisor. It will try to categorize the objects based on similarities, patterns, and differences.

For instance: Different pictures of dogs and cats are put against the machine. It cannot decide which is a dog or a cat. But it can categorize based on similarities, patterns, and differences based on the features of a dog and a cat. Now the supervisor can label one part as a dog and another part as a cat. In this way, the machine learns on its own and makes changes in the algorithm by itself. Thus, it has less accuracy compared to supervised learning.

Application: Data clustering, fraud transaction, etc.

 

How is AI different from ML?

Many times the word 'AI' (Artificial Intelligence) and 'ML' (Machine Learning) sounds similar. But they are not the same. ML is a subset of AI. Artificial Intelligence allows the machine to simulate human behavior. While machine learning permits it to learn from experience. Intelligent Computer systems are made from AI to solve complex problems. Like humans, these systems can perform any task. AI consists of both machine learning and deep learning. The common applications of AI include Siri, online game playing, intelligent humanoid robots, the automotive industry, etc. It aims at doing work at a faster rate without any error.

It is categorized into three types as Weak AI, General AI, and Strong AI. At present, we are working on the first two categories. Strong AI is still beyond the reach of humans. It is more powerful than humans.

 

Applications of AI

Self-driving and parking vehicles: The system working in the vehicle can see, think, learn and move. It will apply the brake when needed and also change the direction. Revolution has already started in the automotive industry a few years back. 

Digital Assistants: Google Assistant, Siri, and Alexa are the digital assistants that a user uses for browsing, checking schedules, and perform daily activities. They are conversational AI bots. 

Vehicle Recognition Identification System: A traffic surveillance camera is integrated with AI technology to read the number plates of the moving vehicle. It helps authorities to search specific number plates. 

Cybersecurity: It assists in recognizing patterns of threats and backtrack the attack. A combination of traditional security techniques and AI can provide 100% results and cut false positives. 

AI against Covid-19: Thermal imaging at airports and other places is possible with the help of AI. It will also track the performance of the medicine on the body and the spread of infection in the lungs. 

Detection of misinformation: AI software will search sensational words that spread misinformation and fake news on social media. They call it 'mining of words'.

 

What is Strong AI?

It is equal to human intelligence, also known as true intelligence. It has characteristics like solving puzzles, planning work, learn and communicate. Experts believe that it will get developed between 2030 and 2045. At first, it will act as a child-like brain and will develop like an adult through learning. Strong AI will be able to communicate with the world by acquiring common sense and languages.

 

Risk of Strong AI

It raises potential threats to humans. Some people fear that it will overcome human power and raise concern over their survival. It will act like a killer robot as seen in the movie I and Robot causing harm to humans.

Another concern is- loss of jobs. It will lead to a massive employment shift. It will replace humans providing more productivity at work. Government has to take the necessary steps to provide safety against the concerns raised by it.

3 Comments

  1. Wow! Thank you! I constantly wanted to write on my site something like that. Can I take a portion of your post to my website? Best Electrical Services Noble Park service provider.

    ReplyDelete
    Replies
    1. yes you can:) share it to me after writing

      Delete
    2. mail me how will you use content of this blog. email id:keyurashiyani95@gmail.com

      Delete
Post a Comment
Previous Post Next Post