How AI is dropping the cost of prediction analytics

Have you ever heard about Moore`s law? In 1965, Gordon Moore, who would later become one of the founders of Intel, wrote a paper claiming that the number of electronic components, which could be placed into an integrated circuit, will double every year. This exponential became known as Moore`s law and turned out to be the foundation of the digital world.

When the semiconductor prices started to fall while their productivity increased, the economy started to use more and more of them. Dishwashing machines, hand watches, cars and of course – computers.

This is what is happening with the economy when a certain resource got cheaper, says Avi Goldfarb, Professor at the University of Toronto. We just start using more of it. It happened with semiconductors and computers. Since the basic task in front of any computer then and now is arithmetic calculations, today we are outsourcing almost any calculations to computers at a greater extent than ever. Games are an arithmetic problem, music is an arithmetic problem, even images can be turned into an arithmetic problem, says Goldfarb.

Why there is such a buzz around AI?

AI (Artificial Intelligence) is not a new conception at all. It was always believed that it had transformative potential. The reason for so much buzz around AI today is the simple fact that just like the semiconductors price fall while their productivity surge, the main benefit of arithmetic calculations are way more accessible. What is the main benefit, you might ask? Welcome the prediction technology!

How do prediction technologies work?

When we talk about predictions, we don’t mean clairvoyance in any form. It is about educated guess, backed with the right set of data. We use historical data to project the values within future data. This way, the prediction is to find a missing information backed with an actual data. Since we think of AI and Machine Learning as prediction technologies, just like the semiconductors and arithmetic calculations, we`re currently seeing the fall of the prediction costs. Hence, the economy will consume more of it.

Just like the insurance you`re paying right now. Its cost is based on already predicted risk, which is evaluated in specific currency and those calculations are made by machines, powered by specific algorithms and some of them have AI accomplishments. Bank loans are the same. The prediction subject here is the creditworthiness of a particular customer. Whether he or she will pay its loan or not. There was a time when humans calculated such predictions but nowadays it is a machine task also known as a scorecard.

Professor Goldfarb says that as prediction gets more and more affordable, the economy finds more and more applications like medical diagnosis, object identification even autonomous driving can be prediction problem.

This is how the cost of prediction have dropped and this is why AI is so hot right now. The lowered cost opened new horizons and anyone is in hurry to discover new prediction applications with disruptive potential in fields, which were inconceivable just a few years ago.

The real value of big data

We feel we are living in a world running out of resources. Except for a single one. Data is the only resource that grows exponentially, says the chief scientist & co-founder of A4E Alexander Efremov, Ph.D. As an experienced data analytics expert, Mr. Efremov is aware of the fact that more and more data will be generated and hopefully, utilized. Let’s say once again that data is the new oil.

What we do with such constant data accumulation, you might ask? Prof. Avi Goldfarb of Toronto University offers an interesting perspective by reminding us that some high profile experts like pilots and MDs are subject of extended learning and training process. To get their licenses, they need to accumulate a spectacular amount of data via professional experience. Their learning process today is making them better tomorrow. Just like Google, which back in the day has invested a piece of their profits into new people with new skills analyzing the searches users do in order to create better search algorithms. By learning, we are getting better and better not matter if we talk about a single person or the entire organization.

data scientists survey

Just like humans and organizations, machines learn from their experience. Say hi to Machine Learning, which is not equal to AI but it is a significant part of it. By having more and more input data, machines are able to learn more and more. The more they learn, the better they will become. Experiencing better machines means improved predictions, which are more accurate than before.

Access to better predictions does not mean that AI is good enough as human intellect. What AI does not have is proper judgment, says Goldfarb. There are decisions, which cannot be made by machines. For instance, even though planes, ships, and cars already have an autopilot function, they cannot reach from point A to point B without people in charge, yet. Maybe someday this will happen but it would not be today, nor even tomorrow.

Would access to permanently improving predictions will result in people replaced by AI? Definitely yes but not all of them. At A4E, we believe that human capital is an expensive resource and we know that when they lose their prediction-tasked jobs, their experience, hence learning capital, should be utilized as creative power.

Leave a Reply

Your email address will not be published. Required fields are marked *