Original article was published on Artificial Intelligence on Medium
Will Robots Ever Take Over The World?
Probably not. Emphasis on probably.
AI, or artificial intelligence, is the latest buzzword in tech news outlets. Not only are robots going to take your job, but it’s racist, too.
Even though a lot of us realize that the internet is always more dramatic than in real life, the AI explosion of recent years does beg some pretty significant questions: How many jobs will be replaced by AI? Is there going to another workplace revolution similar to that of the Industrial Revolution? When will AI begin phasing in and unskilled labor jobs getting replaced?
And although this line of questioning is completely legitimate, innocent, and very well needs to be addressed by our politicians before it arrives at our doorstep, it almost always leads to a form of hysteria and panic, which can lead to the ever-famous question: Will robots ever take over the world?
I’ll give you an answer upfront: the answer is no. Or at least, probably not.
In general, when I refer to robots, I am referring to artificial intelligence, or AI, as a whole. Robotics is a small subset of AI that usually involves some kind of machinery and locomotion.
But what is AI then? Truth be told, it’s difficult to put a solid definition on it. A much better solution is to give some examples of applied AI before moving on to the semantics.
This is probably the most popular AI application. With Elon Musk, founder of Tesla and SpaceX, making headlines practically every day, there’s little doubt that you’ve also probably heard of their autopilot program that his electric car company is developing.
The creation of AI to drive a vehicle is an incredibly complex task. There needs to be route navigation, computer vision, and the ability to deal with uncertainty. Each of these systems requires huge datasets and advanced AI, and to top it all off, they have to work in tandem. Flawlessly.
Never before has AI been so efficient at recognizing the contents of an image, or the creation of new images. There are numerous websites that allow you to look at computer-generated people. Though some of the fake people have some obvious glitches, an overwhelming majority are indistinguishable as being fakes. And, with the increasingly large datasets, the AI used to generate these fake people is becoming more effective. For example: just recently, one of the AI developed for generating fake headshots of people was able to reproduce imperfections, such as moles.
These people are usually dubbed “deep fakes”, a play on the method that the AI uses to learn, called deep learning.
Even scarier, the technology is advancing to be able to generate videos. Several weeks ago, videos showed the AI in action, as a student was able to be Elon Musk in a Zoom meeting.
The impact that this can have on our already bogged-down-with-misinformation political system is unknown, but one can imagine the damage that can come from realistic, faked videos being available to opposing politicians with money to burn.
So, you have some examples. What is AI?
Artificial Intelligence has no formal definition, but can be best defined by looking at some hallmarks of the above examples: AI needs to be able to adapt, and it needs to be autonomous.
Essentially, AI is a program that needs to learn from given datasets and it needs to do so without human intervention.
But Can It Take Over The World?
So you know what AI is. It’s a machine that learns. But that doesn’t quite answer the question.
Though, to best answer the question, I’ll need to go into the history of AI, and then refer back to our examples above to give you a complete, well-informed answer.
When AI was first being developed, all of the buzz was about “General” AI, sometimes referred to as “Strong” AI. This was artificial intelligence that mimicked the human brain: it could learn on its own and do a large set of tasks.
The science fiction authors and directors took this notion and ran with it, making the classic sci-fi movies and vetting the public to think of the dangerous robots that would enslave the human race every time that AI was mentioned in the news.
Though, in the real world, almost no progress was made on constructing a “general” AI, despite years of experimenting and research. Eventually, the idea of artificial intelligence would seem like it would only come to fruition in the realm of science fiction.
All hope was not lost though!
Eventually, researchers began shifting their focus onto “Narrow” AI, sometimes referred to as “Weak” AI. This was AI that could do a very small subset of tasks, and was really good at it. However, “Weak” AI is practically useless outside of it’s intended environment.
This shift of focus from the impossibly complex general AI onto the possible narrow AI has created the recent AI resurgence we have seen in the last few years.
Narrow AI has been created for lots of applications and been implemented much more than the average person realizes.
For example, the self-driving car. The driving of a car is an incredibly complex task as described earlier, but it can be broken down into small subsets of tasks. Each of these small tasks can be effectively handled by narrow AI. There can be AI that chooses the best route to take. The AI that controls the computer vision will be trained to monitor road lines and watch for obstacles. There will also have to be some form of AI to deal with uncertainty, but this could be built into the other systems. The real programs used in self-driving cars are much more complex, but it illustrates how these complex tasks can be broken down in simple tasks that can be handled by narrow AI.
Will general AI ever exist?
This is the million-dollar question.
Although general AI seems confined to the realm of science fiction, there have been some advancements.
Without diving too deep, there is a model of AI, dubbed neural networks, that mocks the human brain’s method of learning and thinking. It is theorized that, with enough computational power, that this model of AI could effectively replicate a brain.
Though, theories are cheap, and we have not even come close to this massive accomplishment. Humans may never be able to replicate their intelligence to the fullest, but that’s not to say that it is impossible!
So, Can AI Take Over The World?
After examining the types of AI and some of the applications, I believe the answer to our burning question is no.
As of right now, general AI is fantasy. Though, the idea of robots that could learn at an incredibly fast rate and perform large amounts of tasks in the same vein as a human is very scary. General AI could very well become our new overlords should it ever be invented.
However, even with the incredible advancements, the route that AI development is going is making it very unlikely that we have to worry about kissing any shiny metal rear-ends anytime soon. Narrow AI is being designed to focus on one task, and are practically useless outside of that task.
So unless the field of general AI makes any advancements, my answer is no. At least, probably not.