While I am writing this article for laymen who may not have the technical knowledge of machine learning and artificial intelligence, I must disclose at the very outset that I have been a researcher myself and used machine learning algorithms for my research which classified the satellite images for the land use – crops, urban land, forests, etc. I have also worked on other kinds of digital images including medical images to find objects. I also have some research publications in conferences and journals to my credit. So, that sets up the authority ! Now, in this blog article, I argue that there is nothing called artificial intelligence and associating intelligence to a machine is at the most a branding or positioning strategy. I do not have any reservations if you use the word ‘machine intelligence’ but by using the adjective ‘artificial’, you are basically comparing the performance of a machine which just understands 1’s and 0’s to a human being whose brain is so mysterious that we still are not sure how it works. And it is not just the brain, we still do not understand the origin of life. We have not yet been able to document all the flora and fauna on this planet. We still do not know what is there in the interior of the earth. We still have not found the remains of the Malaysian Airlines flight MH-370. So, it is nothing short of arrogance to use the word ‘artificial intelligence’.
We see a lot of earth movers today by companies like JCB which can replace hundreds of thousands of manual labour. We see busses by companies like Volvo which have sensors and processors that make many decisions more efficiently than human beings. Do we call these machines intelligent? I do not think so. There are two main types of machine learning algorithms. The first type is know as supervised learning where we have some examples of input-output pairs. This is like the socialisation of a child after birth. We are showing the child something and telling what that must be called or how it must be used. By seeing, feeling and using it again and again, the child understands to differentiate between a tiger and a cow. Even after becoming an adult, if you show him or her some extinct animal, he or she will not be able to classify that animal or name it. Machines using supervised learning also suffer from the same disadvantage. But, human beings can come up with novel ways to handle such uncertainty. Machines cannot! The second type is known as unsupervised learning where we do not have examples but we try to find the structure inherent in the data. What is more intriguing though is to understand the source of this structure or pattern. To understand that source and to solve problems which are social, human being is an indispensable link. Machines can only understand numbers and to interpret these numbers as social constructs needs the inventor of these social constructs – The human being. This is the reason there is only one kind of intelligence and that is human. At best, we can say that the intelligence in machines is nothing but glorified automation resulting from the programming of the machine by a human being.
I am definitely not undermining the utility of ‘artificial’ intelligence. All I am saying is that maybe we must use some other word. Machine learning seems to be apt. I also wish to host a website at some point of time where I will list the tasks that human beings can do and which machines cannot. These tasks can act a sort of challenges which machine learning researchers can use.
Sudhir Gupta's Blog 
I agree that the term machine intelligence is a marketing gimic. But calling it statistic is too general as well. Pattern recognition seem like fair term ?
In current day the machine learning field has grown and everything is not just supervised and unsupervised. There is generative and reinforcement as well.
LikeLiked by 1 person
I think I never called it as a statistic. All I meant was that you cannot compare machine and human intelligence and the word intelligence itself could be a misnomer.
LikeLike