We still have the edge over AI.

AI can be a difficult topic to stomach. You may have experienced a sense of being overshadowed as you learn about all of the tasks we have trained AI to execute flawlessly. Take, for example, chess. In the 1990s, a group of computer scientists at IBM developed a computer-based chess engine capable of considering roughly 200 million moves in a second. This program was named Deep Blue, and in 1996 it was put to the test in a match against Gerry Kasparov, a chess grandmaster and arguably the greatest chess player in the world. At the time, this match seemed like a match between David and Goliath. No chess engine in the 50 years since they were created ever truly rivaled a grandmaster. In fact, the computer-based chess engines were often humiliated as human players ran circles around them. In this match, Kasparov accomplished exactly what we expected and was able to defeat Deep Blue in a tournament that was scored 4-2, but the chess engine had proven itself a worthy opponent. Despite this, IBM was determined to prove the capacity of Deep Blue. They scheduled a rematch against Kasparov in New York that would be broadcast to the world live in 1997 – a bold move. IBM’s team set off to make improvements to the learning algorithm and the hardware that ran Deep Blue. When the time came for the rematch, Deep Blue appeared as a humble yet focused opponent. The game started dramatically as Kasparov took the first win using the “King’s Indian Defense” only to lose the second game by a terrible mistake on his part. The following three games were draws as humans’ greatest chess warrior fought valiantly to defend our undefeated title. In the sixth and final game, Deep Blue stomped out our hopes by beating Kasparov in just 18 moves and winning the tournament. This win over Kasparov marked the beginning of a new era of chess AI’s dominance over humanity. In the last 26 years since that epic battle, no human has been able to get the better of a chess engine in a tournament. Learning of this event sparked a sense of existential dread within me. I extrapolated that this high level of skill AI had demonstrated through chess meant that chess-playing AI was equivalent to human grandmasters in an overarching ability to do tasks. But this was a mistake. Deep Blue can be thought of as extremely narrow-minded. Perhaps, the metaphor of tunnel vision would be more appropriate. Deep Blue and Gerry Kasparov only have one thing in common, playing chess. Deep Blue is extraordinary at playing chess; however, Kasparov is capable of so much more. For example, he can brush his teeth, play tennis, hold an intriguing, meaningful conversation, make art, drive a car, etc. While Deep Blue cannot even play checkers. This difference means the world, Kasparov is able to try a new thing that he has never seen before and interact with it in a meaningful way, whereas Deep Blue can only play chess.

This difference can be understood by looking at the difference between AI and AGI. AI stands for Artificial Intelligence, and AGI for Artificial General Intelligence. Deep Blue is an AI, only capable of doing precisely what it is trained to do. Humans, on the other hand, have General Intelligence. This means that we can use past experiences and learned skills and apply them to something brand new that we have never experienced before. We can perform an enormous variety of tasks constantly and without much effort. To illustrate this point, imagine your neighbor calling you and telling you they will be late coming home from work. They ask if you can put the package sitting on their porch inside their house. You, being a helpful person, agree to perform this task. You think to yourself, no problem, it will take 5 seconds. Let’s focus on this task in steps. To effectively execute this single task, we would need to execute many little tasks. For example, putting your shoes on, locating your neighbor’s house, grabbing the package, opening the door, finding a good place to put the package, etc. For the sake of the point, let’s focus on one step, opening the door to your neighbor’s house. Now, in this example, this is a task you have never performed before. This will be your first time ever entering your neighbor’s house. To a human, this is no problem because of our generalized intelligence. We have handled plenty of doors before, pushes, pulls, handles, and doorknobs. Entering the door code and operating your neighbor’s unique door handle is new to you but doable because of context clues and past experiences. However, if a door-opening AI (paired with a robotic body) was tasked with opening this new door. It would be required to try and fail in everything it could possibly think of to get this door open before finding success. This is because of machine learning, the way an AI learns, essentially trial and error followed by cost-benefit analysis. As it learns, this lengthy and arduous process performed by an AI amounts to something different than the experience we would have. It has created a series of laws that it must obey in order to execute the task ahead of it. An AI wouldn’t necessarily understand what it is doing or why it is doing it, simply that after it performs ‘X, Y, and Z’ tasks, we say good job and allow the AI to quit. Because of AI’s lack of true understanding, it is incapable of “living” in our world.

Artificial Intelligence has yet to cross the threshold into general intelligence. Until it takes this monumental step, humans will continue to be essential to civilization. We will still have our “edge” over artificial intelligence. Some researchers believe AI could bridge the gap to generalized intelligence within this decade, while others argue that we would be lucky to see it happen within the century. Regardless, even when AGI is created, it will be long before they can rival humans in adaptability and task completion. Similar to how it took chess engines 50 years of development before they could rival humans. As for you and me, there is no need to worry about AI coming into this world and overshadowing us now. We can enjoy the beauty of life knowing that we can tie our shoes and play chess.

-James Otteson

Thank you for reading! Please comment with your thoughts and opinions on this topic below.

Deep Blue versus Garry Kasparov. (2023, January 17). In Wikipedia. https://en.wikipedia.org/wiki/Deep_Blue_versus_Garry_Kasparov