Can machines think? This was a fundamental question asked by Alan Turing 70 years back when he broke the Enigma code in WW2 and changed the world by introducing us to computing. Living in the technological age, every person thinks if the devices they carry in their pockets possess intelligence. After all, the smart devices we carry can recognize our voice, and we can summon them to do work for us.
So, after 70 years, every person asks the same question Alan Turing asked: Can machines think, and do they possess intelligence? Well, artificial intelligence answers this question, a science which deals with the intelligence machines possess.
Simply put, artificial intelligence is a branch of computer science which deals with machines that depict human intelligence. This enables them to perform tasks that humans can only do, but now machines can do them.
Self-driving cars are a prime example of what artificial intelligence is; it lets cars drive themselves, which definitely requires human intelligence. That said, artificial intelligence is not about big cars driving themselves; even a set of small swarm robots working together to perform a task are a part of artificial intelligence. Any device, big or small, displaying any form of intelligence comes under artificial intelligence.
Artificial intelligence is a multidisciplinary science, and both machine learning and deep learning are subsets of the same. Now that we have a basic understanding of what AI is, we can look at why we need artificial intelligence in the first place.
Also read: Robotic Process Automation (RPA) Vs Intelligent Process Automation (IPA)
Types of Artificial Intelligence
There are three broad categories of AI, and they are explained in the sections that follow.
- Artificial natural intelligence Refers to machines that can only perform a particular task. In our day-to-day life, we interact with such an AI system. Examples of this kind of AI include voice recognition systems like Alexa, Siri and spam detection features to sort emails.
- Artificial general intelligence: These systems are designed to possess general intelligence and perform several tasks. These systems have not been fully developed yet, but GPT-3 developed by OpenAI comes pretty close to general intelligence systems. You can use GPT-3 for creative writing, program apps and even build advanced chatbots.
- Artificial super intelligence: This domain of Artificial Intelligence refers to systems that will possess more intelligence than humans. Sounds scary, right? People like Elon Musk and Stephen Hawking think the same and have said that such systems can harm humanity. Thankfully systems with super intelligence have not been developed yet, but looking at the advancement in AI, we could see a super-intelligent system in the future.
Also read: TPU vs GPU vs CPU
Why do we need AI?
Let us say you want a machine to perform a particular task, and if you are familiar with how to program it, you could define specific rules it should follow to complete the task. This is the programmatic approach.
If you are using an artificial intelligence approach, you give a machine data and the task it must perform. It then uses machine learning algorithms to understand the data and perform the task by looking at patterns in the data.
No rules are defined for the machine; instead, it uses machine learning to define the rules it needs to perform the task. Therefore, in such a scenario, the machine programs itself making it robust.
This approach helps solve problems that cannot be solved by typical programming. A great example that shows that a programmatic approach can’t solve all problems is spam detection used in emails.
If a programmatic approach were used to make a spam detection system, it would have to be updated constantly as spam mails change rapidly. Programmatically, finding the rules to filter out mails would be impossible.
In an AI approach, the machine would learn how to detect spam emails using machine learning algorithms. In such a case, the machine will detect the change in spam emails and start putting them in the spam folder automatically without human intervention.
So, we need AI because many problems cannot be solved by a programmatic approach and AI intervention is quintessential in some cases.
In addition to all this, AI is making life easier for humans, be it recommendations on Netflix or business intelligence software for big conglomerates. AI is helping us understand the enormous amount of data we generate every day to make our life better.
Also read: How do browsers work? The science behind rendering a webpage
How do machines become intelligent?
Before answering this question, ask yourself, how do humans become intelligent? Well, humans perform a task multiple times, they fail, and after trying several times, they perform the task with perfection. It is here that we say that a human has become intelligent.
A similar pattern is used in machines, but as machines only understand data, the same is used to make them intelligent. A typical AI system takes data from the environment, processes it using machine learning algorithms, and then decides which task it should perform according to its data. This training of a machine using intelligent algorithms, which lets it take decisions, makes them intelligent.
These algorithms use reinforcement learning, neural networks and other machine learning techniques to help machines understand data and take decisions accordingly.
Also read: Dual boot vs Virtual Machine