During the past 2 years, we have seen a bunch of articles and talks in the news & media about the AI and an AI driven society in near future. Most of this atmosphere has been shaped by the success of autonomous car projects and the fact that we would see these cars on the roads around in near future. But the interesting fact is that we have been living in an AI driven society for the past 10 years! Perhaps non-expert people in the field of information technology do not know that most of the information flow on the Internet has been controlling by AI in the past 10 years: All of the search results and advertises on Google, Facebook, Twitter, and the other prominent search engines and social networks are made by AI-based algorithms.
With the advent of social networks and smartphones, everyone has the ability to share everything on the Internet. This ability has originated the formation of billions daily posts by people around the world in recent years. It is quite evident that this huge value of information (the so-called Big Data) cannot be processed by human and we require a lot of autonomous information processing servers to process them. As these types of information are human made (versus machine made), they require some levels of AI to be processed and categorized. This is why we have seen the extensive use of AI by the big Internet companies since 2010.
Now the question is how much successful has AI been in the past 10 years? The answer: Not too much! Perhaps many AI researchers and tech journalists do not agree with me, but I have my own reasons. Recently, there was a Congress hearing in the US regarding some issues with Google company. Among them, one was of more attentions: While you search the word “dumb” in the company’s image search, it shows you the images of the US President! It may be quite fun for you, if you support the President’s opposite party. But it is completely scary, if you are an information technology expert. Why? This is the reason: If the AI-based algorithms can introduce the US President (which is probably the most important person in the world) as a dumb, what can these nice algorithms do with you and me as ordinary citizens?! How can I be categorized by these algorithms? As a poor man on the corner, or a lunatic on the grass, or a geek in the town?! The mentioned sample is an evidence of the fact that AI could be very misleading in some cases which can subsequently bring some serious consequences in different aspects of our social lives including political, commercial, cultural, and mores. It may wrongfully affect the result of an election or the rise and fall of a company’s reputation, and many mores of these bad scenarios, which many of them are not generally got coverage in the news and media.
The disputed Google Image search results, originated by the word "dumb"
The reason behind the above mentioned problem is that what is known as AI in the media and literatures these days, is a set of advanced and trainable categorizing algorithms. These algorithms have got some notable progress in the last 2 decades because of the great advances in the computational power of data centers and some good software tunings including the so-called deep learning. But all of the prominent AI researches know that there is meaningful gap between consciousness and categorization, that is, even a well functioning categorizing system is far behind the capability which is called consciousness. Some researchers believe that the consciousness is the borderline between machines and animals. This is why we call the categorization-based AI as unconscious AI. The problem with unconscious AI is that it can be easily deceived by fake news makers, fake social media accounts, SEO geeks, and human crooks in general. In fact, by injecting some false or manipulated information to some proper places, one can change the information flow in a social media or in the entire Internet! It is really a scary and dangerous situation which needs some serious attentions or perhaps some regulations.
Personally, I do not believe that we should stop the usage of unconscious AI as it is currently impossible to do that due to the share-oriented culture of the today’s Internet users. Instead, we need some standards and perhaps regulations for our current AI-based algorithms. If not, we will see more and more troubles in all aspects of our social and even personal lives which will be resulted by the misinforming of unconscious AI.