One of the biggest tech trends of the last year has been Artificial Intelligence (AI) proliferation. Even though AI has been around in one form or another for what seems like forever, it has recently become much prominent, both in enterprise and in consumer applications. In spite of the numerous ways in which AI is starting to be used, there is one major misconception about AI.
The misconception is that AI either is, or soon will be, superior to human intelligence. In reality however, nothing could be further from the truth. So if it is true that AI can’t even begin to compete with human intellect, then why does AI have a reputation for being smarter than we are?
I don’t think that there is a single answer that can be pinned down as definitively being the reason why AI is so often portrayed as being of superior intelligence. Instead, I think that there are a few different things contributing to AI’s undeserved reputation.
Hollywood has probably played the biggest role in the way that AI is perceived. After all, there is no shortage of movies in which the machines decide that they have had enough of their human overlords and revolt. The “Terminator” movies, the “Matrix” trilogy, and “2001: A Space Odyssey” come immediately to mind. Since I’m one of those people who hardly ever watches television or movies, I am sure there are probably many more examples that I don’t know about. In any case, Hollywood has definitely set some unrealistic expectations of what AI can do.
Another reason why AI has a reputation for being of superior intelligence, is because we have all seen (or at least heard about) really compelling technology demonstrations that show just how smart AI supposedly is. Way back in 1997 for example, an IBM system called Deep Blue world champion chess player Gary Kasparov.
There is little question that there are certain tasks that computers can handle better than humans, but does an ability to win at chess really mean that AI is superior to the human mind? Hardly.
AI systems are like humans in that they make decisions based on the information that they have been given. Like a human however, AI sometimes makes bad decisions or gets the wrong answer. For example, AI-based systems such as Google Home or Amazon’s Echo or Alexa sometimes misunderstand spoken commands. One of the things that distinguished AI from other types of computer programs, of course, is that AI systems are designed to learn from their mistakes.
The reason why AI systems sometimes seem to be smarter than their human counterparts is because computers are really good at quickly parsing large amounts of information. To see why this gives the computer such an advantage, consider a chess match between a human and a computer. In such a match, the human looks at the chess pieces on the board, and contemplates the next move based on the game’s objectives, and based on what they think their opponent is likely to do.
A computer does things differently. Some chess-playing computers are far more advanced than others, but let’s look at a really simple example of how such a system might work. Rather than trying to anticipate what move an opponent is likely to make, a computer might determine every possible move based on the board’s current arrangement, and then use statistical analysis (based on past experience) to determine which move is most likely to yield the most favorable result.
In other words, at least in this example, the computer can’t be absolutely sure of which move it should make. It is only making an educated guess based on past statistics. The more games of chess the computer has played in the past, the more data that the computer will have available for making decisions.
So if that is the way that an AI chess-playing robot works (and again, I am simplifying things), then how is AI not superior to human intellect? I would make the case that there are several pieces of evidence supporting that AI is of inferior intelligence, and to think otherwise is the single biggest misconception about AI.
Superior or inferior?
Let’s go back to the chess-playing robot example, but for the sake of discussion let’s assume that the robot is playing its very first game of chess. As such, it has absolutely no data to go on. The best that a computer can do in this type of situation is to make random guesses as to what might be the best move. Sure, the computer will learn over time, but during that first game the computer might as well be picking its moves by flipping a coin.
Now consider a human who is playing chess for the first time. Let’s suppose that the human knows the rules of the game and the end objective (just as the computer would), but has zero experience with actually playing chess. Even though the human might not have any experience to draw from, they aren’t forced into taking completely random guesses. The human can derive cues from external sources, such as their opponent’s facial expressions. The point is that early on, the human is using true intellect while the computer is taking random guesses. Even later on, the human is still using intellect, while the computer is doing little more than analyzing statistical data.
If you need more evidence that AI is not a replacement for human intelligence, consider that human beings are extremely versatile with regard to their capabilities, whereas today’s AI systems tend to be single purpose. With that in mind, let me give you a really simple example of why even the smartest AI systems are no match for human intellect.
Suppose for a moment that an AI chess-playing robot has played trillions of games and has accumulated so much data that no human has the slightest chance of being able to win against it. With that in mind, imagine that the robot is playing an outdoor match against a human player when it suddenly starts to rain. The human realizes what is happening, and either opens an umbrella or seeks shelter. The AI system, on the other hand, is completely oblivious to the changing conditions and keeps on playing just as it was programmed to.
Of course it would be relatively easy to outfit a robot with a rain sensor, but that’s beside the point. Even if an AI system could mimic all of the human senses (which is technologically possible), it cannot possibly be programmed to recognize every conceivable stimulus and to react accordingly. For a human on the other hand, reacting to stimulus is second nature.
Nothing more than tools
Although AI may be better at certain specific tasks than humans are, AI systems are ultimately nothing more than tools. Tools have one purpose — to extend human capability. I can’t drive a nail into a two-by-four with my bare hand, but a hammer makes the task easy. Just because a hammer has capabilities that I don’t,does not mean that the hammer is going to replace me or become my overlord. Like AI, the hammer is a tool that helps humans complete tasks more efficiently.
Photo credit: Bago Games