What is Artificial Intelligence (AI) and Why should you care?
You must have heard about AI. So what is “Artificial Intelligence (AI)”? To understand, we need first to understand what computers are.
The word “computers” means someone or something that can “compute”. Years ago, before computers come to being, humans who did calculations were called computers. Now computers can carry our calculations, data manipulations, or control equipment connected to computers as per our instructions. Humans can do all those things, but computers can do it faster. The computer can do more than 2–3 billion transactions per second while we can hardly do one. This disparity is evident if we change the time scales. If the computer takes one second to do a complicated calculation, keeping the same speed ratio, it will take a human 63 years to do the same calculation. Also, computers are much cheaper in the long term, compared to humans.
If you want to find documents that have the word “AI” among 10,000 documents or create 20 copies of the same document, computers can do it very fast. Therefore, computers are very good at calculating, storing and sorting through data, and finding relevant information among a lot of data. With these skills, computers have become ubiquitous and have already transformed the world.
However, there are many problems that computers can’t handle. We already said that if we can tell the computer what to do, it can do it. However, with some problems, it is hard to tell what to do.
For example, let’s consider driving a car. Remember how you learned to drive a car. Did someone write down what to do in all situations? If you think you can do it, try to write how to drive. You will find that it is virtually impossible. There are just too many cases. Instead, we learn by trying it out and getting feedback on how we are doing.
A lot of problems that need decisions, complex thinking, or intelligence have the same behavior. AI is a technology that can solve this kind of problems. You do not have to tell AI what to do. Instead, you have to give a lot of examples of what is right and what is wrong, and the technology is capable of learning from it, just like a human.
When AI learns how to solve a problem, it can solve it much faster, again and again. For example, it can recognize the voice, find what is in a picture, diagnose a disease, detect fraud. These things could not be automated because one could not tell a computer how to do them. AI achieve these tasks much faster and cheaper than humans.
For example, let’s say you plan to travel to Japan, but you will find most of the details are in Japanese. You can go to the google translator (https://translate.google.com/) and translate the information. Years back, to do the same, either you have to find someone who knows Japanese or hire a translator. However, now Google can afford to give it to you for free. This is one of the powers of AI.
In the last few years, for some problems, AI got closer to going beyond human performance. The famous example came in 1997 when Deep Blue, the chess-playing computer, defeated the world chess champion. Other examples are Alpha Go defeating world Go champion, and IBM Watson defeating the champion of jeopardy, a natural language quiz competition.
Those examples are from games. Among real-word examples are self-driving cars, Google photo search, language translation, and disease diagnosis. AI is taking over a lot of use cases. It is no longer a tool only at the hand of multi-million companies. For example, about a year ago, an AI program was written by a teenager helped overturning more than 375000 parking tickets. IBM is trying to adopt their Watson AI engine to health, and if successful, it might let everyone get high-quality medical advice for a fraction of the cost.
Furthermore, we interact with AI daily by using applications such as Google search or Facebook. For example, McKinsey has estimated that about 14% of existing jobs will be automated in the next 10–15 years.
Impact of AI or our lives will go much broader than suggested by the above examples. Let’s consider self-driving cars as an example. Let’s examine some of the nonobvious effects. If most cars are self-driving, it will cut down most of the accidents because if one accident happens all cars will learn from the accident. Cars will be able to drive much faster as a single convoy reducing traffic significantly. Most of the parking space can be released as you can get off the car and ask the car to park. Cars will look very different. You can eat, sleep, or watch TV in the car. Traffic police, insurance, and many related services may not be needed.
However, all impacts are not favorable. Some of the jobs will be replaced. People will have to learn new skills. The government has to help people transit between jobs. However, AI does not mean they will always replace humans. Even more often, AI will help humans make better decisions, do things more efficiently, and take over repetitive parts of their jobs.
It is likely that you have seen at least one movie that has killer AI. Indeed that is possible, although we are likely 50–100 years away from that kind of AI. For example, a human can learn from 2–3 examples while AI needs billions.
Finally, before passing judgment on AI, we need to understand the current inefficiencies. For example, many in the world do not have access to top-quality professionals like doctors, teachers, designers, and lawyers. Due to this many die daily. AI has the potential to make professional knowledge ubiquitous and daily save many lives. That means there is a cost of inaction as well. Is it fair to deny AI and let people die or disadvantaged? We have to measure positive impacts against negative ones.
What this means is the world as we know is changing very fast. Our children will live in a different world and do very different jobs. Challenges they face will be very different. Nevertheless, I am hopeful, humans have adapted to many changes environmentally and tech, and we will adapt to this as well.