"It could be terrible and it could be great. It's not clear. But one thing is for sure, we will not control it."
Elon Musk said this about the current rise of artificial intelligence, something he admits to being frightened of and which he believes is evil. He is, however, a proponent of Transhumanism.
Contrast this with Facebook founder Mark Zuckerberg, who sees AI as the future solution to all the world's problems; in reality, his and Facebook's problems.
Current Generation AI
"Go" is regarded as the most complex boardgame in existence. It has been played for 2500 years and is considered the oldest boardgame in the world, still played by millions to this day.
In 2016, a Google-funded project created a robot called "AlphaGo" that was designed to play the game. It faced a human opponent, 18-times Go world champion Lee Sedol, and beat him 4-1.
Go is not a simple game; there are over 10 to the power of 170 moves available to players. That's 10 with 170 zeroes after it. To put that crazy figure into perspective, there are only 10 to the power of 80 atoms in what we know as the observable universe.
Just 1 year later, a new upgraded Go-playing robot called "AlphaGo Zero" played the original AlphaGo robot. The results? It beat the "old" robot 100-0.
What's most amazing, and also deeply troubling, is that the new robot learned how to play without absolutely no human interaction whatsoever. It's secret? By building it this way, it was not restricted by our clearly limited human comprehension.
All that the robot was given were the rules of the game, nothing more. It basically played against itself internally in order to "learn" and became the greatest ever Go player in the game's 2500 year history within just 40 days.
All on it's own, it worked out strategies to beat "us" at our own game. This demonstrates that there are more non-human tactics than there are human ones.
This also suggests that it could apply to literally anything in the world; that's scary.
From this example alone, we can see that the ramifications of an AI device could be far-reaching. And this one was self-taught with only the most minimal data possible built-in.
What would happen if the AI became so "intelligent" by learning and connecting to the Internet that it simply took it over? Is it possible to "switch-off" the Internet? I don't know, but even if it is, the AI could become so advanced that it would already be prepared for that and override any unwanted human interference.
We would find ourselves stuck with AI that is not only more advanced than us, but getting exponentially more intelligent all the time, as was shown in the AlphaGo Zero robot.
What happens if this AI decides to flick the switch on it's creators; humanity?
Artificial Narrow Intelligence
Also know as "weak AI", this is where we are now. This is the only form of AI that we have created so far. Even the Go-playing genius robot is considered as weak AI.
This weak AI is created to perform a sole task; examples of this are seen in factories and production plants around the world. They only do one thing, again just like our Go-playing genius robot.
The likes of Facebook and Amazon use this weak AI to show us recommendations based on our previous browsing and shopping habits. What's worrying is that even this so-called narrow intelligence is able to "learn" and teach itself to be more effective on these platforms.
This is a science based on trying to program AI to learn like humans do. In the same way that a human baby learns over time, so does the AI. But much faster.
The most common method used to make AI mimic a human brain is known as a "Neural Network". This takes the way neurons in our brains fire to transmit information, like an assembly line, and transfer it the best they can into the code of the AI machine.
This means that the AI can not only analyse data given to it, but learn from it and adapt it's own "view" of it. Far from replicating human intelligence, is this starting to sound a bit too uncontrollable for comfort? If so, you are not alone.
The neurons in our brain relay information at around 100 metres per second. Computers transmit information at the speed of light; roughly 300 million metres per second.
A computer can process the same amount of data in 1 week that would take a human brain 20 thousand years.
Artificial General Intelligence
AGI, also known as "strong AI", is simply AI with more than a single purpose like that of the weak AI we have now.
AGI is still in the theoretical stages but the intention is for it be equal to that of human "intelligence". Of course, being equal to humans would just be a starting point.
Fortunately (for most of us), this is proving to be a very difficult task for developers in the field. For example, humans can create and invent things. The difficulty they face lies in the concept of how to teach a computer to create something that doesn't exist or indeed, hasn't even been thought of yet.
If they do manage to solve these issues, due to the exponential growth seen in weak and strong AI, we could see some sort of "Super Intelligence" rise out of this within months or even weeks.
This would see us arrive at the point of what is known as the "Technical Singularity". This will happen when AI becomes so advanced that what it "creates" itself would be beyond the comprehension of humans.
Emphasis on "will" happen, there are no restrictions on what is being attempted in the field of Artificial Intelligence. Theoretically, at this point AI will quickly become billions of times more advanced than it's human equivalent; we will be left behind by "our" own creation.
Intelligence vs Wisdom
There is a big difference between these two. Intelligence is "the ability to acquire and apply knowledge and skills." Wisdom is "the quality of having experience, knowledge, and good judgement..."
Wisdom is far more nuanced than Intelligence. It incorporates insight, perception and judgement. By these definitions, Intelligence can be programmed; Wisdom can not.
Intelligence alone does not include values or morals of any kind. For example, it's easy to see how a computer designed to end world hunger could arrive at the conclusion that the most effective way would be to exterminate most of the earth's population, thus removing the issue. It's this lack of wisdom that is so very dangerous to us.
Seeing as we don't even know what consciousness is yet ourselves, it means that we are not be able to program it into any form of AI just yet.
How would we program something along the lines of our brain's "limbic system" into a computer? This would be necessary to create basic survival instincts like our own; the fear of death would be very useful as a form of control in any advanced AI.
Could AI be programmed to Love? The genuine emotion of love. If so, this creates yet another problem; it would by default also know how to hate.
Does any of this matter? Maybe not to us, but be sure that those working on AI are thinking about these intrinsic human emotions and more right now.
It raises questions as to whether AI could be selfish, greedy, lie and be aggressive, all the so-called negative traits of humans.
What if it were to get in the wrong hands? In fact, can those that desire it and are actively creating it be the "right" hands? They seem to be recklessly dismissive of all the dangers.
The fact remains that the ball is well and truly rolling. Our only hope can be that something blocks it's progress before AI can't be stopped by humans and it goes on to cause widespread damage or destruction of our civilisation.
I believe we, as a species, need to stop searching for perfection, especially in the field of Artificial Intelligence. Perfection is one of the great unknowns, as is the potential of AI.
However, if things go to plan as certain people want, AI will go down as the last invention ever created by humanity.
Did you like this article?
Thank you for your vote!
14 June 2019
Scientists Getting Closer to “Smart Dust” Brain Implants That Communicate Via Wireless Network
From our advertisers
7 hours ago
US planning ‘tactical assault’ on Iran nuclear site to reciprocate 'tanker attack': Report
From our advertisers