Why Machine Learning (ML) is so popular in the 21st Century

The term artificial intelligence (AI) is very topical, but it was invented in 1956 by John McCarthy, Marvin Minsky and Claude Shannon at the famous Dartmouth Conference. It is an old concept that has evolved over time and is related to the idea of building machines capable of thinking like human beings. This characteristic is attributed to Russell and Norvig in 1995 "systems that think like humans".

One of the most rapidly evolving areas of Artificial Intelligence is the machine learning (Machine Learning, ML). Although the previous authors make a distinction between human thinking and acting vs. rational thinking and acting (which could be the subject of several studies), the fact is that the learning attribute can be related to "the human". 

ML is a branch of AI based on the idea that systems can learn from data, identify patterns and make decisions with minimal human intervention. The novelty is that systems learn with little or no supervision.

Technologies and AI solutions are used today in companies for the transformation of current business processes, to boost customer interaction, improve decision making and increase employee productivity. 

But the most interesting thing is to consider that this technology is at the heart of daily consumption: Amazon, Netflix, Gmail, Siri, Alexa, Spotify... The list is huge and all of them are companies at the peak of their value by capitalization. 

The systems recommend millions of users, observe the results of their recommendations and learn for the next recommendation. All in seconds.

Technologies and AI solutions are used today in companies for the transformation of current business processes, to boost customer interaction, improve decision making and increase employee productivity. 

But the most interesting thing is to consider that this technology is at the heart of daily consumption: Amazon, Netflix, Gmail, Siri, Alexa, Spotify... The list is huge and all of them are companies at the peak of their value by capitalization. 

The systems recommend millions of users, observe the results of their recommendations and learn for the next recommendation. All in seconds.

Machine Learning (ML) Popularity Response

The question of why machine learning is so popular in the 21st century can be answered with three types of reasons:

On the one hand, technical reasons: AI and ML are algorithms that work because the technologies that support them have developed exponentially. The massive use of data (BI), storage and processing capacity (Cloud) and the interactivity of the elements of complex systems through common protocols (APIs).

On the other hand, there are reasons of effectiveness and efficiency for companies: the ability to equip themselves with algorithms that learn from their own predictions almost in real time, changes the strategy of business and allows the existence of personalized offers to each customer at an increasingly lower cost.

It sounds paradoxical but the definition of "segments of one" (a defined offer for each customer) is possible because the marginal cost of creating that specific offer for that customer is minimal and the profitability per customer is greatly increased. 

Years ago we would have thought that maximum personalization comes at an ever-increasing cost, and while that's still true for physical products, in the world of consumable services, from insurance to movies, it no longer works that way.

The third reason for the increasing popularity of ML is to be found in the customer. The digital consumer, aware that his data is used to improve your offer or identify a trend, assumes the advantages of handing over his data to ensure a return in the form of greater comfort as a consumer.

The legislations make efforts to ensure customer awareness in the assignment of the use of their data, but the return in the form of comfort that the user gets far outweighs the "potential dangers" that they warn them. Will the companies be able to pay the customer for their data? It already happens. Signing up for a service entails better conditions and data monitoring is accepted.

Netflix, Amazon or Spotify are clear examples of new businesses built on existing content (music, movies or consumer products) but accelerated by the ability to recommend (ML). 

But these big "mainstream" names should not hide from us that technology is also available with great success both in sectors less associated with trends (health, finance, security, for example) and in smaller companies that benefit from the use of technology.

What is the limit of machine learning? 

It is impossible to say here. It does not seem that the technical aspects are going to be a major constraint. Nor are the companies going to be a brake because digital transformation is a fundamental part of the agenda of CEOs and Boards. 

There are undoubtedly limits to the regulatory aspects and the ethical regulation of companies and corporations.

The user or customer himself will be the owner of his ability to "be predicted" and these changes will give rise to debates in universities, companies and regulators. For now, there is a lack of signals identifying these limitations, beyond data protection itself.