How exactly does AI work?
Chances are, you may have heard of Artificial intelligence a lot (even if you don’t have anything to do with it). But have you actually wondered how do they work? I’ll quickly explain 3 terms: Artificial Intelligence, Machine learning, Deep learning.
Artificial Intelligence:
The area of computer science which relates to the ability of computer systems to think and with the reasoning abilities of an average human being.
Machine Learning:
It is a subset of AI and refers to computer algorithms which perform certain specific tasks without using direct instructions but instead, relies on patterns and it’s on inference.
Email providers have been using this for quite some time to distinguish between spam and legitimate mails.
Deep Learning:
It is a subset of machine learning which is capable of learning on it’s own by using unspecified and unlabeled data. It is the most advanced field in AI which brings us really close to realising the dream of machines with thinking capabilities much like humans.
It is commonly used to make programs which can identify cars and road signs from images (I’ll get to this topic in a minute)
How is AI trained?
This is actually a tricky question, if you ask this to some AI specialist or something, they will probably need a moment to think.
It involves providing an algorithm with training data in order for it to learn. With this data, the machine learning model predicts the answer (in most cases) by identifying and using patterns which it then compares to data upon which it has to work.
To illustrate this, let us take a very common example: a machine learning model to distinguish between cats and dogs. For this, we need to keep in mind that the more data is available at our disposal, the better our model will be. So, in this case, photos of cats and dogs are fed to train it.
How do corporations acquire data necessary for all this?
Sometimes, it may be the case that the pictures themselves are not good enough for the model to correctly predict. In such cases, humans may need to intervene.
Companies like Amazon actually pay people just to help them at tricky places, from the comfort of their homes while companies like Google train their models through captcha.
If you remember filling any captcha recently, then the website asks you to select squares which have cars, or maybe even traffic signs. Did you think about how all this data from millions of people is being used? Let me guess, probably not.
All that data (as of now) is being used to train AI for driverless cars (Google Waymo).
YOU have provided all this data for FREE.
How AI is already being used:
Most of us have a smartphone these days, and irrespective of their brand, they usually have some sort of personal digital assistants built in: Apple’s Siri, Google Assistant, Microsoft’s Cortana, and Amazon’s Alexa are all examples of them. They are being run with AI, with the exception of taking the help of humans at times.
As I write this article on Google Docs, the website automatically changed “taking help” in the previous paragraph into “taking the help”.
These examples help us realise how something as futuristic as all this has slowly crept into all our lives without any of us noticing.
Like most things in today’s world, AI too is being involved in all sorts of controversies.
The Pentagon has set up a 10 billion USD contract ‘War Cloud’ which essentially means that they are looking to work with a tech company to develop a military grade cloud-based AI computing system. Currently, Amazon and Microsoft are competing with each other to get this offer.
Why not Google?
Earlier this year, Google pledged not to contribute their efforts towards AI weapons projects amidst employee rallies and demonstrations. Around 10,000 people signed a letter stating that “Google should not be in the business of war”. Subsequently, the CEO, Sundar Pichai wrote in a blog post that he would make sure that Google works towards AI applications that will contribute towards benefiting society and will scrap all projects including project Maven: which was supposed to improve drone strikes by bringing about improvements in facial recognition.
Even companies whose products have nothing to do with technology are using AI to stay at the top of their game.
Coca-Cola is reportedly using machine learning to personalise their offerings across different regions. People in different regions have different preferences when it comes to beverages. So what they are doing is that if someone buys coca cola from vending machines, the machine simply records which options are the most frequently chosen. Using all this data, they are able to cater their drinks according to local people’s preferences. And that is how the people of the US got Cherry Sprite.
Usually, most companies don’t go this far when it comes to leveraging the power of AI. The most common thing companies do is just use AI to figure out their reach on social media platforms.
References: