Hello,
Thank you for visiting our website. You are joining us at an extremely exciting time for AIBODY as we begin our journey to revolutionize medical education. As CTO with a background in artificial intelligence (AI) I hope to use our blog as a medium to share with you some of my knowledge and opinions on hot topics in the AI & healthcare technology world and hopefully instigate some meaningful discussions around these.
My first post is in 2 parts and covers a simple but very important question, why is it crucial to understand what AI is? Part 1 of my post briefly covers the history of AI and sets us up for a discussion of some of the key issues & misconceptions around it in Part 2 next week. Let’s dive in.
What is AI?
Artificial intelligence has many definitions of various complexity. Let us consider the most intuitive and generic one, AI is a program or machine, capable of reproducing a human’s cognitive abilities, such as learning, thinking, knowing, etc.
Do you remember HAL 9000 or Skynet? That is exactly what we are talking about! I am sure that most people associate the concept of AI with these (or similar) fictional systems. This is the main reason of the so-called AI winters, disappointments, misunderstandings, scandals and hype surrounding AI.
A bit of history
During the recent decade AI has become a red-hot topic. Investors are trying hard to spot the AI in the startup’s name or presentation. Top managers are going the extra mile to implement AI in their companies. Meanwhile, scientists all over the world are constantly inventing and implementing new models and methods. The number of articles and research mentioning artificial intelligence, machine and deep learning is increasing exponentially.
You will be surprised to hear however, that the idea for the creation of AI was conceived in the 1950s. That was the very period when the concept of AI was introduced, and the first prototype of modern artificial neural networks was created. As a Ukrainian, I am especially proud of the fact that the inventor of multilayer neural networks and the so-called "Father of Deep Learning" is the Ukrainian professor Ivakhnenko. This extremely ambitious task however did not receive considerable support from the scientific community at that time and one of the reasons of course was the insufficient development of computer technology, which did not allow researchers to effectively implement, verify and improve their ideas.
The current heyday of AI begun relatively recently, in the 2000s. The AI scientific community now even has its own celebrities. One of the brightest personalities in this field is the so-called "Godfather of Deep Learning" Geoffrey Hinton. It was he who breathed new life into neural networks, improving the mechanism of their training - backpropagation. And it was his team, in 2012, that enjoyed a stunning victory in the famous ImageNet Large Scale Visual Recognition Challenge where the AlexNet neural network, created by Hinton’s Ph.D. student Alex Krizhevsky, achieved previously unheard-of record low classification error %. This was revolutionary and marked the start of dramatic growth in deep learning which is a unit of machine learning and deals specifically with artificial neural networks. There is a considerable number of models and solutions emerging every year, and it is stunning how many artificial neural networks are becoming superior to the people who train them. I am talking from personal experience here as last year I spent a considerable amount of time thinking one of my image recognition models was wrong… until I realized the model was simply better than me as it was spotting such details that would be of an extreme difficulty for a person to notice.
But there is a problem… winter may be coming.
The current heyday of AI, due to the rapid development of artificial neural networks, is significantly different from all previous waves. The scientific community now has powerful computing, cloud solutions and access to the vast amount of data provided by the Internet. Clear blue skies ahead right?
So, how could winter be coming? Well, it is a tricky one. It is all about interpretation and perception and I will cover this in more detail in the 2nd part of my post next week. Keep an eye out for it and thank you for reading.
Alex Mokin,
CTO
