Recent studies have shown that the line between artificial intelligence and artificial general intelligence is starting to blur. Artificial general intelligence refers to a computer system capable of performing any intellectual task that a human being can.
And while the distinction is starting to blur, virtually all AI systems are incapable of this level of complexity. As such, artificial intelligence is a computer system capable of performing intellectual tasks. A popular way to differentiate AI from the rest of technology is by using the three laws of thought.
The first law states that “something cannot be both a and not-a in the same sense at the same time”. The second states that “something cannot be more than and less than itself at the same time”. And finally, “if a thing is equal to another thing, it will remain equal if it changes in any way”.
This is known as Aristotelian logic and has been used for over 2000 years. The first law is the law of non-contradiction. The second law is the law of excluded middle. And the third law is the principle of causality. These laws define anything that exists and can not change in a way that goes against these laws. As such, this line between artificial intelligence and artificial general intelligence no longer exists.
Artificial intelligence refers to a computer system that exhibits intelligent behavior, while artificial general intelligence refers to a system that exhibits intelligent behavior on any level of complexity (i.e., some system exhibiting vaguely intelligent behavior in one field could be called an “artificial general intelligence”).
It’s important to note that artificial intelligence and artificial general intelligence are not their own category, but part of the larger “artificial” category. Any sufficiently advanced technology is indistinguishable from magic.
To quote Arthur C. Clark “Any sufficiently advanced technology is indistinguishable from magic”.
Arthur was right, many people still don’t understand how a cell phone works because the level of complexity is too high for them furthermore many people have never even seen a cell phone before let alone know what one does. As an example of how this rule applies in practice, consider a computer driving a car or a video game playing itself.
A computer that can do these things is arguably artificial intelligence. To a greater degree, the computer is an imitation of human behavior. If the computer were capable of writing programs itself and teaching its own “self” more complex things (e.g., programming) it would be evidence of artificial general intelligence.
The distinction between AI and AGI
Moreover, the distinction between AI and AGI is largely irrelevant. The most powerful and cost-effective (and thus relevant) systems are those that operate at the level of AGI. At the same time, systems that operate at the level of AI are still incredibly powerful, especially when dealing with the massive amounts of data they require.
As such, artificial intelligence can be thought of as an umbrella term referring to all such systems that exhibit similar characteristics. It is possible to have a very intelligent system without it being an artificial general intelligence, but an artificial general intelligence is likely to be much more intelligent than one not designed specifically as one.
AI has gotten quite a bit of attention in recent years because it’s getting better and better at doing things that people used to think only people could do. This is only because of the rapidly advancing hardware and software being developed to enable these systems.
The example given at the beginning of this article is that a computer can now beat a grandmaster in chess.
In fact, it’s gotten so good that a computer can beat literally every human player in chess. What makes this feat significant is not only is it nearly impossible for a human to play perfectly, but if you look at the rules of chess you will find no rule stating a player must beat every other player to win.
Chess by itself has no goal or end state (i.e., there isn’t an artificial intelligence system). Rather than win, chess is a game designed specifically to be challenging. As such, humans have designed the rules of chess so that everyone must pass the same test for the system to win (i.e., on all levels of difficulty).
With chess, computers are reaching human-level intelligence in areas where people thought only human intelligence was possible. This is because computers are programmed with goals and specific strategies that allow them to reach this level of intelligence.
Think about an example of something completely abstract in nature: art. Art is something humans have created. It’s impossible to determine what it is without looking at the end result or how that end result was achieved.
The ability of AI and AGI
That is, art doesn’t have an end goal that it attempts to reach.
Artificial intelligence systems now have the ability to create art by learning concepts, rules, and patterns through experience and feedback. Once they’ve learned enough, they’re able to create art of their own. In fact, they are constantly creating new works of art. As such, artificial intelligence is very different from human-level intelligence. Unlike human-level intelligence, it doesn’t have a specific end state and instead learns more complex patterns as it continues to repeat the same process over and over again.
Artificial general intelligence refers to systems that also exhibit some level of general intellect (i.e., systems can put two things together and understand what those things are). Humans are able to do this.
As an example, if you walk into a room with a bunch of people, you can understand what’s going on in the room without anyone explaining it to you. This is because you have an experience that lets you connect the dots.
AI is already at human-level intelligence on certain systems and is advancing exponentially as more hardware and software become available to it.
In conclusion, artificial intelligence is an umbrella term that refers to any system that exhibits intelligent behavior (i.e., computers) while artificial general intelligence refers specifically to a system that can exhibit intelligent behavior on any level of complexity (i.e., computers).