Source: Deep Learning on Medium
AI is not going to kill you! Yet…
Data science and machine learning are two terms that get thrown around a lot nowadays. Most people think of Artificial Intelligence as basically a robot that is going to take over the world and kill most human kind. Hmm? Even pioneers in the tech field such as Elon Musk, point out the dangers of AI. If you’re not too heavily involved in the area, you have no choice but to trust someone who has a self-driving car and a brain computer interface company.
I also would like to point out that the person who is telling us how dangerous AI is, has companies that depend solely on machine learning which is the most fundamental block of AI. I think what Elon Musk is saying when he talks about the “danger” is the “power” of AI instead or it should be.
I’m going to explain why AI is nothing to fear and how it offers a virtually infinite amount of solutions to any given problem.
I have another article about business solutions that AI can provide so if you’re interested feel free to check that out! Let’s start with what AI, machine learning and data science is. Then I’m going to move onto talking about different fields artificial intelligence plays a role in. Lastly, I will highlight the benefits and the potential cons.
Who makes AI? Everything and everyone! I know that sounds vague but to make AI, you need data and data can come from anyone or anything. If I want to get more into the professions that make AI, I would say it is the combination effort of data, software, robotics, hardware and machine learning engineers but most importantly and broadly data scientist. Data science is an interdisciplinary field that uses statistical and scientific methods to get information from a dataset. The statical method is key to what makes a machine learning algorithm work. This information can be used to predict, classify or optimize anything.
Let’s say you have a bunch of pictures of brains of people which neurological disorders and healthy brains. To be able to distinguish which is which you would have to learn the common features of healthy vs sick brains. Let’s say you have a set of 100,000 brain images to study to diagnose. You can’t look at a 100,000 images at a time. I can barely focus on 1 which all the distractions in my life. So here comes AI or Machine learning! A computer is not intricate or cognitively intelligent as a human however computers have tremendous amounts of processing speed. (Next time you hit enter a thousand times on your computer frustrated think of how much faster that machine is compared to any human)The data scientist writes a code to “train” the algorithm on all these images. Almost all information in machine learning algorithms is turned into numbers. Going off of a more visualizable example. Let’s say we have a coat company and sold 12 coats our first year 24 a year ago and 36 this year. How many will you sell next year? 48, easy right!? Now imagine you have over 2000 products that sell different every season every day and every year which different quantities and patterns. Now we have a case for machine learning.
What happens is, if you think of all the data as numbers, the algorithm lays out all the features in a graph. In between there are data points that you would like to know the answer to, demonstrated with question marks. To get the answer, a line is needed to connect the dots together but you can get there a million ways. what the algorithm is just the statistical method that helps you get to that certain ? mark.
The machine learning algorithm is how you get there!
Machine learning is a broader term, deep learning which is more importantly associated with AI. Machine learning methods are include anything basic from regression to SVMs to neural networks. Artificial Neural Networks are the mathematical representation of the action potentials and the communication patterns in human brains. The machine in a way mimics the learning structure of the human brain. What everything essential is, is just statistics. A way of finding a curve to get from one point to the other.
Looking at it from a more fundamental perspective, I find it hard to say that AI will take over in the near future because it’s just statistics however I can say that humans can use machine learning to abuse the power of predictive knowledge. It is happening every day. Is it dangerous to a point that should be feared? Maybe not so much. Yes our privacy is a little violated, yes instagram uses a bunch of machine learning algorithms that get you addicted to the app but is it right to attribute the danger to machines? I don’t think so.
There might be cons of AI but AI has helps better human lives. In many areas such as medicine, genomic research, therapies, transportation and business, machine learning helping companies and individuals improve their lives, products and services. We excelled in genetic and medical/pharmaceutical research thanks to machine learning. Self-driving cars will help minimize car accidents. Any problem that we can’t solve AI can help from answering FAQs for your customers to maybe even one day curing cancer. AI is a tool. It’s all about how human kind chooses to use it.