Arxus Blog
AI and machine learning are increasingly attractive to businesses

More and more industries are beginning to see the potential of artificial intelligence (AI) and machine learning. Deevid De Meyer is co-founder of the AI company Brainjar. He explains why AI is becoming increasingly popular and how you as a company can get started.

The possibilities of AI and machine learning are becoming increasingly attractive to businesses. For example, when processing documents. A new document comes in and you want to extract information from it as a company. That's a fairly repetitive job, but just not repetitive enough to program it. In this case, AI and machine learning can help your organization.

Human AI

For companies, AI is often still a difficult concept. For example, we cannot say in advance what the final product will look like or how well it will perform. AI systems are unfortunately not completely error-free, just as a human is not. The tool can learn to perform a task, but cannot explain why it sometimes makes mistakes. This is described in the industry as the "black box problem". To avoid problems, we always recommend that customers have someone physically check the system.


The AI system does perform all the tasks, but an employee reviews everything and approves or disapproves certain tasks. In this way the tool also receives constant feedback, and it can improve and adapt itself to changed circumstances. When everything is still checked by a person, we call this human AI.

Ethical AI

Another form of AI, is ethical AI. It is an extension of human AI. But while with human AI a person checks everything again, ethical AI goes a step further. A good example is a self-driving car. All systems run completely independently, as do the ethical issues that may arise and for which humans have no ready-made answer. Suppose a car in an accident has to choose whether it would hit a mother with child or an elderly person. Here you would not be able to rely on human input anyway, since people also make different choices among themselves.


In many domains we want to make the leap from semi-automated systems, where people can still intervene, to fully automated systems such as ethical AI. But then we also have to start asking ourselves additional questions. An AI system learns from data, from what we put into it, and from certain tasks. There is a notorious example of a large parcel service. It had installed a recruitment tool with an AI component for technical profiles. The tool was supposed to scan resumes, and pick out the most suitable profiles. The system selected mostly men for the technical roles, while there were also suitable women in the mix. It had unwittingly picked up on the preferences of real recruiters, some of which the recruiters were not even aware of themselves. Such a bias can always creep into a system.

First steps

As a company, do you want to start working with AI or machine learning? My first tip is to start looking at your processes. Do you have a stream of documents every day that you want to extract information? Then you can use AI or machine learning for that. Just keep in mind that the tool will need hundreds, if not thousands of examples before it can do the job properly, and even then you'll need human control.

So in many cases there is not 100 percent automation. You first look at which task takes a lot of time and is performed often. Once you have done that, it is best to visit a specialist such as Brainjar. Together we go over the task you want to automate, and look for the right solution for you and your company. Achieve your initial success there, and then you can build to automate more and more processes.

Related Posts

Learnia helps companies train and mentor employees

Digital skills and the digital skills gap are two concept that are very present in the job market today. Digital skills have become indispensable, both in business and in society. Jan Pintelon is Managing Partner at Learnia, and a specialist in learning digital skills. He elaborates on the theme.