Can and Can’t of Artificial Intelligence
What’s the most trending topic in the Industry? Artificial Intelligence. What’s the next revolution in the field of technology? Definitely, Artificial Intelligence. And guess what, according to many, Artificial Intelligence is also the next Industrial Revolution (automated world).
But what actually is Artificial Intelligence?
According to a study by McKinsey Global Institute, AI is estimated to create an additional 13 trillion US dollars of value annually by the year 2030.
The simulation of human intelligence processes and automation by machines, especially computer systems is Artificial Intelligence. In other words, to make machine smart enough and intelligent to overcome the daily tasks by automation, AI is used.
AI is the new electricity!
There is no doubt about the fact that we’re absolutely witnessing the rise of AI. And that goes without saying that it has massively influenced our lives. From Apple Siri, Google Assistant, Alexa to AI Applications in farming or to a huge leap of self-driving car; all of this couldn’t have possible without the invent of AI.
Looking at the future, I don’t think there will be any industry which do not influenced or impacted by the Artificial Intelligence in the next several years.
And when I say this, I do not mean to give you a false hope or to fill a fake bubble in your mind because there are some myths about the role of AI in our life which can be true in the upcoming years but are far from reality at least as of now.
So to clear the unclear, one must need to know what Artificial Intelligence can and cant’t…
What Artificial Intelligence Can and Can’t?
Well, we all know there is a lot of excitement about AI regarding the future but no to forget so as the unnecessary hype about it.
One of the main reason behind it is that we need to understand AI on the basis of two separate ideas. Artificial Narrow Intelligence (ANI) and Artificial General Intelligence (AGI).
There is a lot of work currently going in the field of Artificial Narrow Intelligence or we can even say that almost all the progress we are seeing as AI is happening in the field of ANI (Artificial Narrow Intelligence). Mentioned above applications such as smart speaker and a self driving car are perfect examples of ANI.
But the other type, refers as the Artificial General Intelligence (AGI) is the most valuable type/branch of AI. That is the main goal to build Artificial Intelligence — so a machine can do what human can do and maybe with a touch of super intelligence and do more than a human can.
We are seeing so much progress in ANI, Artificial Narrow Intelligence but at the same time, almost no progress in AGI which Artificial Intelligence.
Looking at the smartest and intelligent products around us, it goes without any doubt that We are seeing so much progress in ANI, Artificial Narrow Intelligence, but at the same time, almost no progress in AGI which is Artificial Intelligence.
The point here isn’t about rating the effectiveness of one over another, but to remind the fact that both are worthy goals. But unfortunately, the rapid progress in ANI which is again incredibly valuable, has caused the people to conclude that there’s a lot of progress in AI, which is true to some extent — but also people falsely think that there might be a lot of progress in AGI as well which is creating unnecessary hype to some irrational fears about the robots coming to take over the humanity anytime now.
Artificial Intelligence is a threat to humanity ~ Elon Musk
Well one can’t negate the quote of the greatest living inventor so easily — but AGI is still an exciting goal for the researchers to work-on. And to set up a platform for the robots to take over the humanity, it’ll require a great technological breakthrough before we get there.
Assuming how far away AGI is, I think there is no need to unduly worry about it. It may take a decade or hundred years or thousand or even an year to get that technological breakthrough.
AI is Machine Learning. But what is Machine Learning?
The rise of AI has been largely been driven my multiple tools — one of those is Machine Learning. The most commonly used type of Machine Learning is a type of AI that learns A to B (a.k.a) input to output mappings.
Let’s say, if the input A is an email and the input B one is spam or not, resulting value could be 0 or 1. Then we are sure that this is the core piece of AI to make a spam filter.
More examples like, if you want to create a chat bot, so when a customer sends you a message with the name of the certain product or ask the price — input A, your bot replies automatically with the best suitable answer — Output B.
This set of AI is called supervised learning which learns input A to output B, or A to B mappings.
The Rise of AI with Neural Networks
The concept of neural network emerges from the neurons in human brain — striving to make machines a lot smarter and enabling the ability to act like a human being brain do.
Deep learning is the advanced or upper level of Machine Learning, comprising of neural networks that trains to perform modern AI, where you feed them more data, performance keeps getting better for much longer.
Thanks to the modern AI and the rise of the Internet, which has given great importance to the data sometimes you hear as ‘Big Data’. Remember, having more data always help you in AI!
So to achieve the best possible levels of performance of your applications, you need two certain things: One is, lots of data that helps to train your neural networks in an effective manner. The second thing is, you want to be able to train a very large neural network. You must require fast computer’s including Moore’s law, but also the latest specialized processors such as Graphics Processing Units (GPUs).
Probably I’m sounding boring to you know? No. Well that’s enough for you to discover about AI. Let’s finish the things in a formal way!
What Artificial Intelligence Can’t?
Is there anything that AI can’t do? Or not possible with AI? I assume this question has been raised in your mind at anytime reading this story. So let’s clear the unclear…
From the past example under the machine learning heading, we’ve seen how we make or train machines through A to B mapping. In retrospect, let’s say we’ve built a facial recognition system that identifies and matches the picture and provide the result in the form of true or false. So this is what we can do with machine learning or deep learning at large.
And now here’s an example what AI can’t do, or at least would be very difficult using today’s AI.
Let’s suppose there is a construction worker standing in your way holding out a hand to ask your car to stop. Also, here’s a hitchhiker trying to wave over a car and a motorbike rider too raising the left-hand to indicate that they want to turn left.
So now, if you’re trying to build the system to learn the A to B mapping, and to train the system to act according to the human’s intentions, that today is a very difficult task to do. Because in this sense, we’re short with data. Part of the problem is that the number of ways people could gesture is very, very large. So yeah, this is where we all need to be careful.
To summarize all this, what AI can do and what AI can’t do; here’s the simple explanation.
The today’s Artificial Intelligence can do everything, that a human mind can think or process to be happened in a second of mental thought.
Thanks for reading, I hope you have enjoyed reading it.
If you are looking for any digital service, be it video animation or website design, or any other, you are always welcome at Logo Sight — Top Digital Company With A Sheer Taste of Creativity.