AI replaces "Big Data" as the hottest buzzword

Just a few years ago, everybody talked about "big data". The managers talked about how valuable data can be and that data should be kept and accumulated as much as possible. The developers, data engineers and DBAs talked about embracing big data platforms (mainly NoSQL platforms and Hadoop).

Recently, an old-new buzzword re-emerged and actually replaced "big data" as the hottest buzzword in the industry, and that is "Artificial intelligence (AI)". All kinds of software vendors strictly label their product with the "AI" buzzword, regardless of whether the product really has true artificial intelligence in it or not. It simply helps the sales because when people think about AI, they imagine sci-fi movies, robots and futuristic innovations.

AI? Really?

Apparently, there is no "intelligence meter" that can measure the intelligence of a system. It seems that when we talk about AI, we actually refer to systems that can “learn” from experience, and for a computer system - experience equals data. The way a system can learn from its data is by extracting insights from the accumulated data and/or from the real time data that keeps coming.

Systems that can learn implement statistical methods and models that in the 1990s were called “data mining” and today are called “machine learning” (ML). ML leverages data mining by constantly updating the prediction models with new data to get a more accurate and up-to-date predictions. The latest developments in ML are in the field of Deep Learning (DL) which takes ML one step further by using multi layered artificial neural networks to search for patterns of data. DL drives sophisticated systems for face recognition, speech recognition, audio processing and video processing (particularly Convolutional Neural Network - CNN). DL also drives systems for natural language processing (NLP) (particularly Recurrent Neural Networks - RNN) and enables the computers to understand texts and whole paragraphs.

How come AI is the current buzzword?

If ML is not really new, how come it is so big now?

1. No expensive software - There are a lot of open source free libraries for developing ML applications which are mature and well documented. ML is easy to use than ever.

2. No expensive hardware - No need to buy and install a tower of computers. We can simply use cloud services with ML built-in. We can easily harness an enormous amount of computing power for building prediction models.

3. We've got the fuel - We’ve got the necessary data on our “big data” platforms to fuel the ML learning algorithms.

4. Simple ML - The cloud vendors (Microsoft, Amazon, Google) offer easy to use frameworks that make ML accessible and easy for developers and data engineers. You don't have to be a data scientist to run simple ML tasks just like you don't have to be a DBA to perform simple CRUD operations on the database.

AI everywhere in everything

Just a few years ago, everybody talked about "big data" and here we are, a few years later, and we see NoSQL platforms everywhere, residing next to the good ol' relational platforms. They have matured and became a critical part of a vast amount of production systems and data warehouses. Today “big data” is everywhere.

The two cloud giants, Microsoft and Amazon, have announced their joint project "Gluon" which might become the leading AI platform in the cloud. This will probably make the "AI-everywhere" vision a reality sooner than expected.

Having “Big data” (the former buzzword) the fuel for AI (the current buzzword), I just wonder what will be the next big buzzword after AI…