Business Talk: Big Data, AI, Analytics: From the hype to business value

“Don’t be afraid of Big data”, if Thierry had one message to share with us during his last business talk that would be this one. Data, Big data, deep learning, analytics, cognitive artificial intelligence,… those words sound barbaric to you? Well, luckily our professor, Thierry, who also runs his own company specialized in AI, is here to decode them. And you will see that with clear explanations, there is nothing to fear.

1) Big data, datafication, analytics, what does it means?

Before giving the methods or secrets about data analysis, better to define the subject: Big data.

What is big data? What are the differences between data and big data ?

Big data is simply a big amount of data, no more. The feeling that it’s something much more complicated is probably linked to the fact that we need to organize those data. The challenge here is then to capture a maximum of data, manage them and find the process to use them.

An other particularity of Big data is that not only they are big but also they are collected in real time. Google, twitter as well as the public surveillance cameras are examples of massive data streams. The power of those institutions is huge but as mentioned before, data itself is useless without any organization and interpretation. Big data can’t be use alone.

What is Data-fication?

“Everything we do leaves a trace in the world.” reminds us Thierry. Indeed, each of our action is recorded and can be used by its owner. Companies understood it; and those with a core business operation relying on data and its infrastructure are progressively transforming into data-driven enterprise. We call this change “Data-fication”.

What is analytics? How far can the analysis go?

“Analytics is a mindset that focus on making sense out of data to have actionable business insights.” stated the presentation of Thierry. In other words, data alone is irrelevant. We need to transform our data into information and make those information a knowledge adapted to our business.

To show us how far can go the analysis, Thierry exposed us a typical data science problem: Is it possible to use telematic data to identify a driver signature? Imagine a car, in this car a driver, but how to recognise without being physically present who is driving? This question can be solved thank to a strong analysis which requires to set features and several variables.

Depending on the difficulty of the problematics, the analysis process is more or less important. Some challenges like “how to recognize a face without label” can be solve with machine learning programs. Machine learning is a second level of artificial intelligence, it consists in memorizing specifics features of one element and then be able to identify it in a different situation.

Today, deep learning breakthroughs the AI world. This expanding technology offers high computing performance. In the field of pictures identification, here is how it works: a huge amount of random pictures are gathered. Among them; several pictures of cats. With powerful algorithms collecting and analyzing different information the process is able to define by himself the concept of cat. And no human action interfered in this operation…

2) What are the challenges?

In 2015, the deep learning; big amount of data collected on several years; enables to create the “Rule game changer”. The computer keeps collecting data and use pattern matching action. To play chess, the computer will analyze your actions and use them to strengthen its game. But create a big database is not that easy…

Although the best level of technology today, deep learning is the hardest to reach. The necessary big data set demands energy, time and methodology. Besides, humanity is infinitely complex and despite the rise of artificial intelligence(1), scientists have to train computers as a proper human.

“Teaching pragmatism to computer is one of the most arduous tasks for a scientist” stated Thierry.

Indeed, human societies got rules, and rules are not always clear and obvious. All the more our cultures differ from where you live in the world. A clear example will help you to understand the complexity of the task: When we say, “I don’t have 100$” , it implies that I have less than 100$, but for a computer, it means that you can have less or more than 100$. Scientists have a lot of work before programming an artificial intelligence capable to distinguish these shades.

(1)Artificial intelligence is the technology that we can find on robots for example. More clever than a computer, a robot working with AI is part of the world. He can perceive, understand and think.

3) How AI can help your business?

AI can help you to make better decisions for sure. But before calling on a specialist to analyze your data, you need to ask yourself the good questions. Is my database consistent enough? How many years of data do I have ? Indeed, if you have less than 3 years of data, a predictive model will be simply useless. 3 or 4 years are the minimum to elaborate a relevant strategy.

Now, let’s suppose you have enough data, what can you do with them? Here are various business axes that can be improved thanks Artificial intelligence:

– Products :Enhancing of the features, functions and performance; creation of new products.

– Operation: Optimization of internal business operations and external processes like marketing and sales.

– Work force: free up workers to be more creative by automating tasks.

– Business shares: capture and apply scarce knowledges to pursue new markets.

In a nutshell, AI brings automation and new inputs. Such competitive advantages can help you to enter new business or take the lead on your competitors.

The offer and demand for AI-driven products will increase rapidly in 5 years time. Indeed, because of AI, offers and internal processes will rise up from 20% to 60% for technology and consumers. Financial service, health care will be also strongly modify by 2023 (15% to 55%).

But today, less than the half of companies are using AI in their process. Some on them are currently on pilot project (23%); others use AI in some of their processes (18%), and for a minority AI is extensively incorporated (5%). So, it’s time to choose your side! There may be an opportunity for your business!