- On March 15, 2019
- 0 Comments
There are many areas in companies, in different industries, that generate a lot of data, and without a doubt today they are being transformed by information technologies to provide business intelligence. It is evident to think that all productive areas within an organization are the first ones that are affected because that is where the value of innovation, growth, and finally, the business of the companies is perceived.
However, there are many other areas within the companies that can produce value for an organization, not so direct or so tangible, but ultimately value. Some companies obviously start with the financial areas, those that manage the economic part of the companies, but very few today do it in one of the areas with the greatest indirect impact with which any organization can count, and that area is precisely the Human Resources (HR) group.
Today in HR everyone talks about skills and profiles. The question we could ask ourselves is: What personality, competencies and skills, numerically, perform better in X or Y placed within the organization ?.
What real data does a company have today to answer this question?
In Nugo, we start from the principle that says that in order to make informed decisions, we need data. So the first task that we face is to create a repository of information about everything that happens in the work life of a collaborator. With this, we can then do information analysis, obtain comparisons, get patterns of behavior and performance, correlations between different variables, etc. This is the first step that today is taken to be able to innovate, to have a lot of data, of many things in an accessible place. Basically, this is what is called Big Data.
Surely some fan of BigData will say that I am hyper-simplifying the point, and in fact, I am. The reason is very easy: this text is not designed for geeks who think about tera, peta, or zettabytes of information, with data-lakes scattered and having to make data transformations in order to homogenize data and meanings, so that they can do multivariate variance models, or things like that (we still do not get there). Going back to our simple Big Data example, if an HR team could store all the information of what happens in a collaborator’s professional life, the courses he has taken, his performance, the notes of his supervisors, his incidences, one, four, ten years; for its 300, 700, 3200 employees; and we could associate these data or relate them to other 3, 427 or thousands of companies, then, the BigData model would begin to make sense for many more, and they would not see it as a simplified model.
And now, with all those millions and millions of data bytes from my collaborators, what can we do?
Introduction to artificial intelligence for HR people
When one has at hand a lot of data there comes ideas of what to do with them. For example, let’s imagine that we have the behavior of wage increases and bonuses paid to millions of employees in the last 10 years, from different companies and industries, from a whole country. Obviously, the first thing that comes to mind is statistics! This information will help us understand the past, but, and here comes the change: How can I use all that information to predict the future?
To answer this, let’s make a parenthesis and see what is, in a phrase, artificial intelligence. We can ask Wikipedia or many other places, but basically, I summarize artificial intelligence in “that a computer can do what a person does”. Solve a problem, analyze information, recognize patterns of behavior, play a game, and a million etc.
Going back to our example, with all that salary increase information, can I determine how much salary will increase in the next performance evaluation period? Surely I will have a number that by probability is close, but how can I get it close enough to say that I can predict it? For this, I will undoubtedly have to add more information to the model in order to be more accurate in my prediction. For example, I will need data on the country’s economy, or growth data in industries, and much more information.
For example, if we exercise our exercise towards the world of recruitment, much of this information could help us determine which candidate will be the best collaborator for a vacancy we have available.
In order to create these models outlined in the previous paragraphs, the people dedicated to creating these artificial intelligence algorithms now have an endless number of tools, many of them with less than 3 or 4 years of creation, with which they can work and improve significantly the effectiveness of their models.
Depending on what you want to achieve or the purpose of what you intend to achieve, there are different areas, approaches or specialties within the world of “artificial intelligence”. For example, to achieve predictive models, predict the price of something, or, as in our example, the percentage of expected salary increase for this next evaluation period, we can use Bayesian networks (probabilistic inferences) of what has happened previously and with some degree of probability, determine what will happen next.
Some other models could use algorithms that learn over time (better known as machine-learning (ML), a branch of artificial intelligence). This is undoubtedly one of the branches that have grown most recently, for its applicability, and for the success, it has achieved in many recommendation systems. I’m sure Netflix or Spotify will get your attention because every time they “know your tastes better”.
In the sub-world of ML, there are many ways that an algorithm can learn, with supervised models (someone is telling the algorithm how well or how badly it is doing it), or non-supervised models (where one gives to the algorithm a mode of self-evaluation). And the way to create the algorithms can be simple linear equations obtained with multivariate regression models, or sophisticated neural networks that are learning according to the “rewards” or “punishments” received by the answers they have been created, and achieve what is called “deep knowledge”.
And, how do I apply it to my model?
In many ways. For example, companies like EmpleosTI, where I currently work as Operations and Strategy Director, uses two different branches of artificial intelligence for their candidate recommendation system. Natural language processing (or NLP for its acronym in English) for reading and interpreting candidates’ curricula, and a supervised ML model for their recommendations of similar candidates. Other companies such as ScreenIT uses ML models based on similarity to make candidates’ recommendations in its application called Zourcing. Mya Systems, uses NLP for its bot, through which it interviews candidates to obtain information and from it create a profile.
All these are exercises that use artificial intelligence only for recruitment and selection of candidates however, the areas of applicability of artificial intelligence for human resources is infinite. Now I have a question so that a system like Nugo could answer us: If I count on Pedro as a collaborator of the company (and I have all his labor information in the organization), and I want (or intend to) put him in the line of succession of Finance and Administrator Director: What are the courses he must take? When should he take them? What positions should he has to have before getting a Director position? How long must he be in each position? What is his probability to achieve it? … and the most interesting of all… What would his performance be in 3 years in the Administration and Finance Department?
Is this our near future? (and, how close?)
About the Author- Eduardo Pierdant.
-Partner Director at Nugo.mx – Talent Management Platform
-Operations and Strategy Director at EmpleosTI
Previously, Eduardo Pierdant was the Technology and Product Director of OCCMundial.com for almost 10 years. During his participation in the largest job pool in Mexico, he was responsible for the strategic, creative and innovative enabling of information technologies on the site that is visited by more than 20 million people a year. During this time it was possible to increase the size of the business five-fold, introduce 3 new products and services, and update the business model to achieve the expected growth. Previously he was Manager of Competitive Strategies at Microsoft Mexico where he held other positions as Information Technology Security Manager. Prior to Microsoft, he worked for companies such as Digital Equipment, offering solutions for the financial sector.
Eduardo Pierdant graduated from ITAM with a degree in Computer Engineering and has a diploma in Corporate Finance from the same institute. In 2004 he certified in marketing at ITESM. He is passionate about his family and sports.