Introduction
The term "Artificial Intelligence" (AI) has been somewhat popular in this era. As a matter of precision, I may be moved to infer that it is one of the mostly discussed emerging technologies in this decade - well, except you've been hibernated all through this era, but if not, then you might have probably had an insight on what AI is. But as it were, there are more things to AI than is just seen peripherally. In this survey, we will be taking a peep into another stage of the evolution of Artificial Intelligence; which is the Artificial Super Intelligence.
[Image Source: Pixabay. CC0 licensed]
To make things clearer a little; especially for the uninitiated; I would give an overall breakdown of the concept of Artificial Intelligence, and the types, then you would see where this tech is potentially driving at. Just so that you may know; and to reiterate again; the idea of Artificial Intelligence is to design a system that would simulate or mimic the human level of intelligence cum cognitive abilities in machines. Just like the name suggests: "Artificial" (meaning; something not natural, or something made to emulate a natural thing) and "Intelligence" (which talks about cognition and knowledge), so putting both of them together, you can figure out the idea behind AI; at least to a considerably fair extent. But the human level of cognition is somewhat complicated - it is not like a kind of textbook that you can just flip the pages to grab any info you require.
Someone could ask; wassup with AI, and has the true idea behind the establishment of AI been fully reached? Okay look at this: from the time the term "Artificial Intelligence" was coined in 1955 by John McCarthy [ref], advances have been made into this regards, but these have been majorly hinged on just the cognitive aspect of human intelligence; but more often than not, other aspects like "common sense", "emotions", and even "logical reasoning" have rather been kept in the gray zone. To understand this a little better, I'll pass us through the stages of how the technology of Artificial Intelligence has been evolving; starting with the simplest form.
Reaction Machines [Artificial Narrow Intelligence]
This is the simplest form of the Artificial Intelligence. You wouldn't be wrong to call it "Artificial Intelligence in the ground state"; and this was one of the first attempts made by mankind to recreate intelligence artificially to produce smart machines. But in this type of Artificial Intelligence, some obvious features are lacking; like the relation of past event to the present occurrences - In order words, the reactive machine is not necessarily designed to initiate a course of action, but to respond to certain situations/events based on some pre-programmed algorithms and codes.
This also means that they are designed to interact with the surrounding/immediate environment and perform some stipulated actions in relation to the event. No wonder the name "reactive" is affixed to them - because, obviously, they are reactive in nature. Even though this represents a major breakthrough in the field of AI, but does it qualify to be called Artificial Intelligence in real sense? - judging from the fact that it is devoid of thoughts, reasoning, emotions, and the ability to form new memories or relating inferentially with an existing database to form new concept or initiate new line of action. It is just like an AI that is straight-jacketed to perform just a single function. I'm sure you're piqued to know the types of Artificial Intelligence that are grouped as reactive machines. Have you heard of the Google DeepMind's AlphaGo? This is an automated Go platform that competed alongside human Go players, and awesomely, it defeated Ke Jie (who was the world's Go champion). This was achieved by systematically analyzing all the possible moves at any stage (so by inference; it was a step ahead of the human player) [ref]. Brilliant you may say, but here's a bit disheartening fact; the AlphaGo wasn't even aware that the game of Go is actually a type of board game [ref]. Damn! Isn't that a dumbass AI? lol. The reason being that; it obviously lacked common sense - it only has one function: Play the damn Go and win.
[AlphaGo Vs Human. Source Flickr. Author: Buster Benson. CC BY-SA 2.0 Licensed]
Limited Memory Artificial Intelligence
Remember what I told you that the reactive machines cannot create an entirely new set of memory; but in this type of AI, the machines can actually form new memories, and can also relate with databases to solve a problem in real-time. Just like the "car collision detection system" in the autonomous-driving vehicles; when it is faced with an impending collision, it would relate with the database on how to circumvent the obstacle, or it can create or chart some new access routes to prevent collision. Also included in this category is the popular personal assistant; Siri, and even some chatbots.
But no doubt, this is nothing compared to the human brain capacity. Just to make it clearer, the human brain; on an estimated average; can store somewhat of 2.5petabytes of data [ref], not to mention that fact that they are composed of neurons that transmit electrical impulses and signals - so you see, the complexity of the human brain is way beyond this. But overtly, this type of AI can be enhanced to form what is known as "Artificial General Intelligence" (AGI). This is the kind of AI that can function in more than one specific task - I mean, they are not stereotypically limited to a particular function like the reactive machines.
[Image Source; Pixabay. CC0 licensed]
In the book of one of the proponents of emerging technology on Artificial Intelligence; Ray Kurzweil - Age Of Spiritual Machines; he referred to the Artificial General Intelligence as the "Full AI", which is also called "Strong AI"- reasons being that it is closer to the human level of cognition. But obviously, this also lacks some qualities, like; self-awareness and consciousness. But there is another futuristic type, which represents the ultimate destination of the technology of Artificial Intelligence - this is the Artificial Super Intelligence (ASI).
Artificial Super Intelligence
Before proceeding; it is worthy to reiterate that this type of AI has not yet been achieved; it currently exists in theory. But again; that something is theorized does not mean it is fictional or not attainable. At that point, AI would have evolved with full self-awareness and consciousness, and would have superseded the human-level of cognition. Not just that; even common sense, thoughts, emotions, rational thinking, feelings must have been incorporated into these machines. Remember the words of another futurist and AI proponent; Ian Pearson:
Artificial Intelligence would be billion times smarted than humans. - Ian Pearson [source]
This is actually a practical definition of what he meant by that. Some would ask; is it possible to achieve this level of Artificial Intelligence, and how close are we to achieve it? Take a look at this: During the earlier introduction of computers, the processing power was just some few Calculations Per Second (CPS), but within few years, this went into billions of CPS. But the human brain (which is what is being simulated) has an average processing power of 1exaFlop (1 quintillion CPS) [ref]. Now that's pretty huge.
But what steps have been taken in this? It might interest you to know that some Cray supercomputers have arisen; like the K computer (which is able to execute 10quadrillion CPS), and the beast Tianhe-2 (Milkyway-2) with a processing power of 33quadrillion CPS. Does this give you any insight of how fast technological advancement is? No wonder the futurist; Ray Kurzweil has proposed that an era is approaching in which the entire human cognitive ability will be processed in apposition with the computational capabilities of machines. This era would be known as the era of "Technological Singularity".
[Countdown to singularity. Source: Wikimedia Commons. Author: Ray Kurzweil. CC BY 1.0 Licensed]
According to Ray Kurzweil, this is set to be triggered in the years 2045 [ref], but this is not where the scares lie, but the effect it would have on the human community. Have you ever imagined having machines that would out-smart you in "every "ramification? But, how da heck do humans hope to build this type of AI? Maybe we don't; but with the designing of Artificial Intelligence that can program other Artificial Intelligence, we could place the creative powers in these machines, and they would chart their own course. Just like in evolution; they can employ all possible combinations of algorithms over-time until they evolve with super intelligence (though this process may even be faster than the normal biological evolution; but it is, however, not expected to happen overnight).
So how would humans hope to control and prevent these AIs from turning against humans? Remember, it would be many times smarter than the average humans; and by inference; this has placed humans as "inferior" to them. No wonder, the late Stephen hawking said this:
Artificial Intelligence could end the human race - Stephen Hawking [source]
But looking at it fairly, if this works out in our favour, then it would be the greatest of all man's inventions, because it can proffer solutions to problems beyond the human ability; and yes; maybe proffer solutions for the reversal of the effects of climate change and many more. So you see, there are two sides to this; but we should keep an open mind and hope for the best.
Conclusion
What is the true meaning of "Artificial Intelligence", and have we achieved this yet? Maybe not, but the potentials would be fully attained with the unleashing of the Artificial Super Intelligence. And the truth is; with the astronomical growth of AI, this could be just some few years away.
Thanks for reading
References for further reading
- Artificial General Intelligence
- Pathway to singularity
- Computational_intelligence.pdf
- Computer Consciousness
- Superintelligence and singularity
All images are CC Licensed and are linked to their sources
