As we usher in new century, we entered a new world-a world of artificial intelligence. Though people were working on AI from last 50 years or so but only now we are able to see practical application and fast development in this area. AI is the simulation of intelligent behaviour in computers using programming. That means AI imitates human intelligence without emotional interference. Since the programming is based on lot of research data AI can perform better than human. AI machines beaten human in games like chess and Go. Silently they are entering our daily life. If you use iPhone you must be familiar with siri- an AI application.
Artificial Intelligence is still in developing stage. Yet many Peoples find it dangerous for human kind. Their intuitions might be right. There are two ways AI research can proceed in future. One is called slow takeoff and other is fast takeoff. A slow takeoff is a situation in which AI goes from sub-human to human to superhuman intelligence very gradually. We will have enough time to study dangers associated with it and to remove those dangers. But in fast take off in few years we will develop super artificial intelligence and that will not give us enough time to study, understand and restrict dangers associated with it. We can not predict behaviour of anything which is more intelligent than us. Presently all data suggest medium to fast takeoff because all big companies and developed countries are spending billions of dollars on it. Within ten years we might reached singularity a term used for AI reaching human brain power.
Machine learning is another major part of AI. Learning without any kind of supervision requires an ability to identify patterns in streams of inputs. It is much more like how a baby learns to interact with world. I find it more powerful than AI based on programming every known thing. Because in that case you cannot expect more than what you input and intelligence of machine is dependent on programmer intelligence. If programmer is not careful enough, machine will make mistake. But in machine learning, based on sensors machine learn itself from its surrounding. It gives machine ability to learn from data without programming it explicitly. This process is slow but has more potential in long-term. We might be able to make a machine which can answer some of the ultimate questions which we can’t answer because of our limited capabilities. Google Deep mind uses reinforcement learning and thus it can improve itself. Alpha Go made by Google deep mind division already beat human in Chinese game of Go which is much difficult than chess.
to be continued in Part-II