Whenever there is a buzzword, and the complex subject it is usually good to start with the definition. It is a little tricky because the definition of artificial intelligence seems to be continually moving. Whenever we solve a problem, we don’t call it AI anymore. It started with chess. A lot of smart researchers looked at other intelligent people. Also, said we are good with math and logic and playing elaborate games like chess. So they started working on those kinds of problems thinking that once they solve them a lot of other things will fall into place. However, they didn’t quite. Because those were simulated environments that didn’t have the same kind of noise we have in the real world. So now researchers shifted largely from playing games which are still an important area and can teach you some things the things that we didn’t consider as that much of high intelligence. Just understanding spoken words seems relatively simple, and we can all do it. However, that was a severe problem up until 2010 when deep learning changed it and was able to make much more progress on this. Now we don’t call it AI anymore its just Siri. Just a speech recognition software. But that was a severe problem that we weren’t able to solve, and there are still some tricky issues in research in it.
Another area that deep learning has made a tremendous amount of progress in recent years in computer vision. Mainly image classification. One of the most imp ideas of AI is called an end to end trainable models. Where we take raw input for instance pixel of imagine and predict the final output. Whether its a cat or a dog, house or ice cream in that image. So as we put that raw input into this models, they keep trying to learn more and more complex representations. As they go layer into a layer, they combine the blobs, edges into a complex texture. As they go further and deeper into different segments, they will identify object parts and eventually merge the elements to recognize the full object.
These are some of the great applications that we are now able to do as long as we have enough data about a particular domain. in fact
One of the most exciting manifestations of human intelligence is language. Moreover, language is making a massive amount of progress right now, but still, there are many ways to go. In 2011, when Forbes realized whenever Anne Hathaway starred in a movie, the reviews came out, all of a sudden the stocks for the company Berkshire Hathaway go up a significant amount. It was already evident in 2011 people were trying to use natural language processing for algorithmic trading and in this case, made the mistake of so-called entity disambiguation they disambiguated Hathaway to the company instead of the actress and then made pretty substantial monetary decisions.
There are some active areas of research that we still work on, one of them is text summarization. Pretty much every natural language processing model you see in the past only can generate at most a sentence coherently. When we try as a community to generate more extended sequences fully automatically in this end to end deep learning models, which usually didn’t do very well.
The next generation of AI will be adaptive, self-learning and intuitive. So when things change that’s when automation kind of fails but the next generation of automation, what’s going to happen is, when something changes it will be able to change its own rules.
Eventually, these computers will be so intelligent that it leads to the singularity, and that means that the human race as we know it becomes obsolete. Exciting isn’t it? So many people think this is what’s going to happen when that happens, but the actuality is this has happened before. 200 years ago 90% of the people worked in the agriculture world. Now 2% of people work in agriculture. Now are we better off or are we worse off? Things got better, how the same thing is going to happen in the coming years. As the technologies get faster and faster, we are going to shift to something else, but it is going to be good for the human race in general because the robots are going to be working for us.
Moreover, the AI world will be virtual, most people see AI as a robot, but really when you google search millions of algorithms are running in the background on servers somewhere else. It is going to be the same idea, but AI is going to be doing tests in a virtual world. We can call it the matrix or whatever we want but what is going to happen is that if you want to find a cure for cancer, what more effective way to do it than a test on a simulated human being a billion times with a particular drug. Instead of doing it on a rat or a monkey, we are going to do it in a simulated environment a million times and let’s say this AI is looking for a cancer drug, and the other is looking for a Parkinson drug. This Ai develops a theory on trying to find a cure for that disease, and it may find something that helps out with something.
40 years from now we will look back to today and say, “we cannot believe we used to do that.” In every field possible. In the next 30 years, there’s going to be just as much change happened that happened in the last 2000. The technology is exploding, but it is going to be good for all of us, we’re actually going to benefit from it and might even be able to work less. We used to work for 80 hours a week, and now we work for 40.