From the introductory part of this post, you would discover that I dropped a question in the last part. To answer that question, we need to re-focus on the idea behind Artificial Intelligence. As the name suggests; Artificial Intelligence is meant to simulate and mimic the human-level of cognition, efferent intelligence, relational intelligence, etc, in machines; but have we attained this yet? I remember many years ago, while I was watching the Terminator movie series, in one of the scenes, terminator wanted to self-destruct himself, and little John Connor embraced him with tears rolling down his cheek. But humorously enough, he didn't understand what tears means, because, clearly he was a machine; which is devoid of consciousness and emotions. And the truth is; humans have been known to be efferent beings, with consciousness and emotions, so why hasn't these been replicated in machines yet? But before we look at this, let's take a review of the advances that have been made in the field of AI, and how it has evolved at every stage.

Stage 1: Artificial Narrow Intelligence

This has been recognized as the basic and first stage of the evolution of Artificial Intelligence. No wonder this has been termed by many as "Reactive machines". The reasons being simple; they are not capable of forming new memories on their own, but are based on a pre-defined (mostly hard-wired) memory. So for every task, they would relate to their database and follow a sequential order to perform the task. Even though this would be considered as the ground state of the AIs, but this was a major breakthrough, because it charted the course for mankind in the advancements of AI technologies we have today.

Human playing against AlphaGo. From Flickr Licence:CC BY-SA 2.0

More so, the limited memory AI can only relate to the present course of action, but cannot initiate one itself - that is the reason behind their "reactive" and not "proactive" nature, because they can only respond/react to actions to perform tasks. Now you might be piqued to know what kind of AI would be classified under the Artificial Narrow Intelligence. Have you ever heard of "AlphaGo"? Okay for those that have not already known, it is an AI designed by Google's DeepMind to play the game of "Go", and that is the only thing it does - to play the game, and cannot multitask. Awesomely enough, it also beat human champions by relating to every of the potential moves in the game of Go. But on a very funny note, the AI was not even aware that "Go" is actually sub-categorized as a class of boardgame according to this source. Well, that is because they are only reactive in nature. People have been thinking that "common sense" is not common in humans, but it is completely devoid in this class of AI. Okay let's look at the next stage and see the advances that were made.

Stage 2: Limited Memory AI

In this stage of AI evolution, one of the major breakthroughs that happened was to leverage on the major weakness of the Reactive Machines (which is the lack of the ability to form new memories) and enhance them. In the Limited Memory AI, there is a pre-defined ability to communicate with a certain database to perform a task, and here, the AI can also make new memories through the combination of algorithms as against the Reactive Machines that are straitjacketed to performing a single and monotonous task/function. Example of this includes your popular chatbot; Siri.

Even though this AI could create entirely new memories, but just like the name suggests, the memory is somewhat very limited and cannot; in anyway; be compared to what the brain capacity of humans can take. So has a true AI been achieved yet? You wanna know the capacity of the human brain? Okay it might interest you to know that your brain has the ability of storing up an excess of 2.5petabytes of data (okay that's a little tip for you). Also, this AI is at the borderline to the "AGI" (Artificial General Intelligence) - which is very much known in our day-to-day lived, which can multitask; like your humanoid robots (e.g: Sophia - which was granted citizenship of Saudi Arabia). Here's a little funny video of Will Smith trying to kiss Sophia the robot, and her response.

Stage X : Artificial Super Intelligence

Okay I named this "stage X" because this particular AI only exists currently in theory, and that is what can be said to be the true definition of AI, because it has been speculated that it would emerge with every of the efferent domains of the human intelligence; like consciousness, emotions, self-awareness, relational thoughts, even way beyond the human level of cognition. Many years ago, a very renowned proponent of Artificial Intelligence (Ian Pearson) made a revelation in this article that AIs would be billions of times smarter than what the ability of the average humans can handle. This type of AI was clearly what he was talking about.

The question is; is this type of AI attainable, and how soon would it be? Okay let's look at our computing world: In the recent past, the processing power of the fastest and most powerful computing systems then would only perform some few CPS (calculations per second), but imagine what is happening now - we have come supercomputing monsters that can perform an excess of 33 quadrillion calculations in just one second like the Tianhe-2. Though the humans brain is way over that processing power; maybe we'll be looking at some quintillion calculations per second. But amazingly, within just few years of technological advancements, we were able to achieve processing power of billions per second, imagine what the world would look like in the next few years - obviously, we would hit the target of quintillion calculations in seconds, and that is when the true AI would be birthed; the Artificial Super Intelligence. But until then, technology would keep advancing.

Conclusion

Since the advent of the first ever Artificial Intelligence, we have witnessed some mind-blowing advances in the field of AI, but the question that has always been kept in the grey zone is if we have actually achieved what could be called a true AI. But one thing I know is; whatever happens, tech always rules.

References

• Advances in computation intelligence.pdf
• Achieving compute consciousness
• Artificial General Intelligence
• Consciousness in machines