IBM Thomas J Watson Research Center
If one follows Information Technology (IT) or business news regularly, one would have come across terms like Cognitive, Cognitive Computing and Artificial Intelligence (AI) today. But exactly what do they mean? Why should one care? What should one be careful about when considering AI? And what is not cognitive? Is it only relevant to industrialized countries or is it an opportunity for a developing country like India?
Motivated by frequent questions from many with technical Computer Science (CS) and non-CS, but scientific, background alike, this article puts forward one perspective to clear the air and help others navigate the technical and business literature. As follow-up reading, one can refer to more technical accounts on the topic [40, 26, 11, 12].
At the outset, here are some definitions.
- Cognitive: of, relating to, or involving conscious mental activities (such as thinking, understanding, learning, and remembering) 
- Cognitive computing: is the simulation of human thought processes in a computerized model. 
- Artificial intelligence (AI) is the intelligence exhibited by machines or software. It is also the name of the academic field of study, which studies how to create computers and computer software that are capable of intelligent behavior. 
The cognitive story from a computational point-of-view began at least from the 1950s when scientists and engineers began to ponder how to make smart machines. They had already built calculators successfully to crunch large numbers during World War II, and had realized that computation, which underlies information processing, had tremendous potential to impact everyday lives. The biggest evidence of smartness around is the human brain and we all can think using it (exhibit intelligence), interact with others (social skills) and act independently (express autonomy). So, could we build machines that would behave similarly? Hence started the fascination for cognitive.
But knowing about thinking is one thing and building systems that seem to think for all practical purposes is quite another. The community divided into the cognitive computing branch, which was concerned with understanding and simulating human thought processes, and AI, which was concerned with building useful, seemingly smart, computer implementations (see sub-areas explained here ). A number of results in those days showed that one could build smart systems without deep understanding of how human thinking works. For example, the Eliza system  could engage people in conversations using shallow rules about human languages. There was tremendous hype about useful systems that may come soon and then people realized the challenges, both technical and of social implications, in building robust intelligent systems.
The two communities have made tremendous progress over the elapsed years and have started coming together again as evidenced in tracks and papers at top AI conferences like Association for Advancement of AI (AAAI) and International Joint Conference on AI (IJCAI). Hence, we will use cognitive and AI terms interchangeably hereafter. However, has anything really changed over the past nearly seven decades? Yes, and no.
State-of-the-art Artificial Intelligence (AI) and data management techniques have been demonstrated to process large volumes of noisy data to extract meaningful patterns and drive decisions in diverse applications ranging from space exploration (NASA’s Curiosity), game shows (IBM’s Watson in Jeopardy™) and even consumer products (Apple’s SIRI™ voice-recognition). But much more needs to be done as the case of Tay chatbot from Microsoft  showed which was in the news in March 2016. Tay was trying to engage people with hopefully better algorithms than Eliza using new cognitive understanding and inadvertently caused social grief because it over-relied on user’s inputs and could not detect that its responses were getting manipulated . Further, none of recent AI systems have provably helped yet beat more mundane and real world challenges that the author considers benchmark for successful usage of computation like fighting diseases, eliminating hunger, making precious water available to the thirsty, improving commuting to work, or reducing financial frauds and corruption. Hence, we have a long way to go.
In the next section, we will review major components of an AI System. Then, in the following section, we will characterize cognitive systems based on their ability to take independent decisions. We will next discuss two case studies of how AI may impact the world with self-driving cars and human wellness, and their social contexts. Finally, we will conclude with a call to action.