What most people know about artificial intelligence comes from brief glimpses in movies or on TV. For example, you may have seen IBM’s Watson, a supercomputer that uses artificial intelligence, appear on Jeopardy a few years ago against two of the most successful human contestants in Jeopardy history. Notice how we had to clarify that they were the two most successful “human” contestants in Jeopardy history?
As you might have guessed, Watson won by a comfortable margin. Even though the other two contestants competed early on, by the end of the two-day game Watson won $77,147 while the next closest contestant only had $24,000.
There’s a catch though. This game of Jeopardy took place in 2011 and the actual Watson computer had to be stored in a different room and have its answers played into the studio because the cooling system that fanned the ten server racks that powered Watson was so loud it would have disrupted the show, so while this game may have seemed like it was the beginning of the end of humanity, in reality it was more a victory that relied on particularly favorable conditions and questions for the supercomputer to even function.
That was 2011 though. What does artificial intelligence look like today and how is it being used?
First, you must understand what exactly artificial intelligence is. For starters, artificial intelligence has nothing to do with self-awareness or emotions. That is sentience. Artificial intelligence can be broadly defined as a set of methods that allow a computer to do something that a human would usually do, such as learn.
While this may sound groundbreaking, artificial intelligence has been around for so long that the original roadblock to advancing computers’ abilities using it was that the computers could not access enough data to analyze and make decisions. The internet changed that, and as it grew more data became available for computers to analyze. Still, computers then were only presenting several possible answers to a problem without actually choosing the correct one.
Later in the 1990s computer scientists programmed computers to learn from patterns in the data they were analyzing and infer decisions based on those patterns, otherwise known as machine learning. For example, if you upload pictures to Facebook the site will recognize the faces of the people in the picture and recommend tagging them based off data gathered from previous pictures, but this is only scratching the surface of what’s being done today with artificial intelligence and machine learning. From navigation apps like Waze to paywalls that decide how much access a person gets based on how likely they are to subscribe, machine learning and artificial intelligence have kicked open the door to limitless applications that could apply to every industry.
Still, the cutting edge is being pushed even further thanks to neural networks and deep learning.
A neural network is a computer’s internal network that is made up of many digital neurons that all relay signals concerning data. For example, if a computer needed to identify your face in a picture of several people each neuron within the neural network would pick a reoccurring feature and learn to identify it. When shown a photo they would find the feature and relay that data to the next neuron until each on has done the same thing and the computer can decide whether you are in the photo or not. The thing that makes neural networks exciting is that they do all of this without being told to. This process of learning without given instructions or parameters is known as deep learning and it has to potential to affect everything from the way we conserve electricity to how our cars are driven.
A computer that simply needs to be fed data and will learn on its own without specific rules or logic programmed into it, is the closest we’ve been able to step towards creating something that resembles the human brain in how it works as well as what it is capable of. As AI, machine learning, and neural networks advance, the possibility of computers that can safely drive cars, alert officials of possible cyber and physical attacks, and much more are becoming realistic.
To the average person, Watson’s appearance on Jeopardy looked like what the first computers looked like decades ago. Just as the first computers seemed like powerful calculators that took up an entire room, Watson seems like just a powerful search engine that needs its own power plant to operate. However, those first computers are what led to the technology we rely on every day now, and if Watson is a stepping stone towards something as impactful as what those first computers created, then there are no limits to what the future holds.