Artificial Intelligence and Machine Learning -Techbytes by Jeremy Howard
Artificial Intelligence and Machine Learning
Programming is considered as a very tedious task by many people because it requires laying down detailed instructions to be followed by the computer, and that too in machine language. The conventional software engineering lifecycle involves planning the approach on a whiteboard and designing algorithms (steps to be taken) before writing the code.
But, what if you need to program something that even you’re not aware of? Deep learning is the answer to this question. In this Ted event held in Brussels, Jeremy Howard discusses the implications of artificial intelligence.
Where did it Start : Artificial Intelligence and Machine Learning
Although the roots of AI date back to the time of Alan Turning and the second world war, the real revolution was started in the year 1956 by Arthur Samuel. Arthur started making a program for playing checkers, which eventually defeated the champion of Connecticut in 1962.
He designed the program in such a way that it learned from hundreds of checker games. It laid the foundational principles of neural networks and deep learning, which are the main pillars of modern-day AI. Since then, researchers and programmers have developed many efficient algorithms which help in computer vision, recommendation engines, audio processing, chatbots, automation, and many other tasks.
How did Machine Learning Evolve
Machine learning was just a stepping stone in this vast domain. Computer scientists are paying huge attention to deep reinforcement learning. This concept allows programs to learn without any kind of training data or predefined instructions. Reinforcement learning is the underlying principle behind self-driving cars.
Machines are beating humans in many things with deep learning, which mimics the working of the human mind at crude levels. These algorithms are designed by stacking nodes and layers, which is analogous to neurons in the human brain. It gave birth to concepts such as CNN, RNN, GAN, LSTM, etc. which are nothing but modifications of the most primitive deep learning algorithms.
Implications of Intelligent Computers
Besides computer vision, text-to-speech & recommendation systems, ‘self’ learning computers have reduced human efforts to a large extent. Moreover, computers are way faster than an average human being and can do repetitive tasks with higher precision.
It is one of the best implications of computers that can learn. The task, which took several hours and had to be done by a bunch of people, can be now completed in a matter of minutes. It saves money as well as human resources for a company.
Artificial intelligence has partially replaced humans, which involved redundant actions. Moreover, with the current rate at which it is evolving, it seems that AI will replace almost every job in the coming years.
Should You Be Worried because AI is Taking over Jobs?
It is not a matter of concern because companies will still need people to govern AI or to crunch data and extract insights. Furthermore, AI still needs to be developed. It requires large teams of researchers. So, there's no need to worry about losing your job.
However, individuals doing jobs requiring manual labor might need to upgrade their skill set towards a more technical and cognitively challenging side. Therefore, we must make subtle changes in existing systems so that we can adjust to AI.
