How does Artificial Intelligence generate employment?

⇒ Biography

Hubert Wasser is a data analysis engineer. After spending 10 years at ESIEA (Paris Engineering School) teaching algorithm and machine learning, he is now Head of Data Science at AB Tasty.

According to Hubert, machine learning (or automatic learning) solicits both fear and hope. Could the frightening future we see in sci-fi films become a reality as we enter into the early days of the machine learning era? Or perhaps, taking a less Manichean stance, does this revolutionary technology offer humankind new possibilities?

As an expert on the subject, Hubert wanted to share how, when put to good use, artificial intelligence (A.I.) can be used to generate business and job opportunities.

When data makes unsolvable problems, solvable.

 

Machine learning consists of using huge quantities of data together with learning algorithms, making unsolvable problems, solvable. One of the best known examples of this is Watson, the IBM machine intelligence, which is already competing against great doctors in the medical space. Another example is Google’s automatic car, which is already on the verge of revolutionizing the transportation industry.

Machine learning is actually more common than you may think, and already impacts our daily lives. Some everyday examples: predictive analysis of a shopping cart values, estimates of risks when applying for bank loans, and so on. Machine learning is, in fact, at the helm of the “New Economy” mastodon development model, (think Uber or Airbnb.)

Machine learning, two distinct approaches:

 

  1. The supervised approach:

In this approach, the learning examples are made up of data and the expected results. For example, to create a vocal identification system, the data consists of voice samples and the expected results are their owners. This approach thus seeks to reproduce expert human responses, learned through the example database.

  1. The non-supervised approach:

For the non-supervised approach, the examples consist only of data, without the expected results. The “learning” is done in the same way for both approaches. In layman's terms, the program decides by itself how it will treat and segment the data, (e.g. the arrangement of faces by sex, age, eye color, etc.) This approach solicits a more profound level of expertise than the supervised approach. Its purpose is to identify new avenues we would never have thought of.

Typically, when it comes to web data, creating visitor segments using the non-supervised approach has many advantages. It allows you to go beyond traditional ideas and detects new behavior or uses.

How do business and jobs come into to play?

 

The supervised machine learning approach can be considered the more reassuring or “safe” form of A.I., as it is restrained and controlled. However, when used, it also has the risk of making people (doing certain jobs) obsolete. Once programmed by the expert, the machine becomes capable of replacing the human being previously doing the task at hand. For example, vocal recognition replaces a secretary for note taking, an autonomous vehicle makes a driver unnecessary, and so on.

When using the non-supervised approach, on the other hand, the role of an expert is to create a learning protocol without a precise objective. This in and of itself leads to smart learning. It’s primary objective is not to automate a task, but rather, discover new ones. In this case, machine learning is not used to substitute a human job. It merely creates information that did not previously exist.

Due to its simple and relatively risk-free nature, the supervised approach is the form of A.I. most frequently used. The major downside of this is of course, the potential to increase unemployment rates. In order to preserve jobs and even create new professions based on new machine learnings, many argue for more frequent use of the unsupervised approach.

Read the whole article on LesEchos.fr .

Related Posts


Tweet
Share
Share
Pocket
Buffer
X