Whether we are talking about AI or ethical AI, it is necessary to define this concept, because there are many uncertainties around this innovation, mainly due to lack of understanding and because of past experiences. The human being fears what he does not understand and in newspapers, films, AI is often used in a dystopian future which can worry the general public.
It is therefore normal that there are “prejudices” and questions surrounding this technology applied to human resources. What if AI was discriminating? How can we avoid recreating biases with technology? Will AI replace humans in HR processes?
AI, the trump card of the HR manager
Thanks to HR technologies equipped with artificial intelligence, recruiters can identify candidate profiles – both internal and external – thanks to skills matching. In addition, AI systems can now help managers understand and detect the skills of employees and candidates in the company, in order to identify and recommend personalized training and career development opportunities. However, with process automation, HR managers fear that the algorithm will discriminate like current traditional processes and approaches do. Indeed, used on a large scale and especially in human resources, AI will have an impact on a large number of people.
Developing ethical AI applications is therefore necessary in order to correct existing prejudices. Nevertheless, it is difficult to define when this technology is sufficiently ethical. One of the main challenges is to transform ethical concepts into practical applications of AI. Achieving this requires a good understanding of the algorithm so that HR can spot any issues while using it. To do this, AI applications need to be analyzed regularly to ensure that algorithms do not learn badly or repeat incorrect patterns.
Ethics through data
In order to create an ethical framework for AI and HR applications, data quality is of the utmost importance. Ditto for understanding the provenance of said data. Companies must therefore use “clean” data because, since the algorithm does only 10 to 20% of the work, it cannot do better than the information at its disposal. Indeed, the previous algorithms were designed from historical data which could therefore repeat old models.
Since then, the European Commission has made progress on the subject and defined seven essential elements (security, protection of privacy, etc.) for ethical AI in order to provide scientists with data to make ethical applications of AI operational. It also announced, a few days ago, its proposal for a reliable AI with an approach based on risk levels. According to the Commission, AI systems in HR present a high risk and they will have to meet strict obligations. Long before this announcement, European companies had already started to take up this issue. In France, Orange announced the creation of a data and AI ethics board to support the implementation of ethical principles governing the use of AI and data. It’s a safe bet that this is just the beginning.
In order to allay the fears around AI, it is essential to educate individuals on what this technology is, its purpose and why we must use it so that they can better understand it. The same is true for its integration into HR processes. It is no longer a dogma but a reality. Of course, we are still only at the beginnings of its development, but we know that this innovation will revolutionize the industry, so it is up to us to lay the foundations for healthy use. It’s still time !
José Rodriguez, Chief DPO, Cornerstone
Expert opinions are published under the full responsibility of their authors and in no way engage the responsibility of the editorial staff of L’Usine Nouvelle.