As of March 1st, 4.5 million Americans used Amazon Echo regularly. “Home Artificial Intelligence”, Echo is also known by the familiar name of Alexa. Although not in the human form, it was recently noted that Echo was able to understand and perform 3000 tasks. In other words, 3000 micro-skills that, just yesterday, would require human intervention. Today, Echo is just the tip of the artificial intelligence iceberg, a set of practices that promises to profoundly disrupt the 21st century industrial economy.
The first impact of automation is to change the notion of performance itself. This happens in two ways: at first, automation allows new ways of measuring that transform our perception of performance. We have new data, so we can know new things, and transform human behavior.
In a second step, the automation of intelligence carries a performance in itself; it can contribute directly to the creation of economic value. It replaces humans, regardless of their behavior.
Let’s take the first case.
Two researchers from the University of Liverpool applied machine learning techniques to analyzing the performance of soccer players. From the simple observation of the movement of the players on the ground, the algorithm was able to determine the distance run by each player, the intensity of each movement, and even the motives behind these movements. Feint, defensive withdrawal, strategic withdrawal, direct attack, indirect, receiving pass… all from a simple visual analysis!
In addition to carrying the promise of annihilating the entire sport statistics sector, the algorithm ranks players according to their value, according to their contribution to the probability of a team winning. To the surprise of the researchers, the “best” players were consistently not those who scored the most goals.
This story is reminiscent of the Oakland A’s, cleverly told in the movie Moneyball, where a baseball team uses data analysis to optimize performance, potential and market value to build a winning team. A practice that, as recalled by the American scientist Scott Anthony, is now widespread in this area.
Yet, how many directors today are able to pilot such statistical studies? How many companies, big and small, have started recruiting a first data scientist to do this?
The main difficulty of this transformation towards artificial intelligence lies most often in a lack of knowledge – and even, in some cases, a disinterested skepticism – within the management teams.
We do not know what it is. We do not know where to start. It feels like it’s a fashion.
Really, we do not understand much.
From the outset, it is difficult to recruit good candidates for senior positions in artificial intelligence. Labor is rare and inexperienced since the sector is relatively new. It is also dangerous to recruit for such jobs without knowing what distinguishes a good mathematician from a bad one.
Secondly, it has been shown that organizations, especially their management committees, tend to recruit people who are similar to them. Under the pretext of strengthening the organizational culture and working more effectively, companies and their leaders are depriving themselves of quality individuals who could probably disrupt their traditional patterns of thinking.
Unfortunately, as many industries have learned at their expense in recent years, the leadership techniques and reflexes of the past do not guarantee future success. However, the transformation of performance threatens those who have risen to managerial positions on the premise of a performance evaluation system itself challenged by the AI.
The difficulty of taking these transformations into account is also much greater for established organizations than for emerging organizations, since they do not have a historical performance measurement system. New businesses develop theirs within the parameters of the data economy and do not have to deal with the cultural transformation of leaders.
The shortage of manpower, combined with the obstacles raised above, nevertheless pushes for more investment in this sector. Beyond the performance of the company, it is the performance of the investment itself that will grow.
Do you already have your data analysis team? If so, all the better. If not, how much do you value the opportunities you let sleep in the meantime?