Vincent Maillard (VM): Today, we are focusing on the use of generative AI in Human Resources. Grégory, this is truly your area of ​​expertise at Danone. To begin, what is generative AI and what are its strengths and limitations?

Grégory Chocoloff (GC): Many people think that AI is a recent phenomenon, but in reality, its origins date back to the 1950s, in American universities. The term “artificial intelligence” was even invented for marketing purposes to obtain funding, playing on the appeal and fear it could arouse. Today, it is interesting to see that we continue to capitalize on this effect to promote new products. So AI has been around for a long time, whether in industry, commerce, video games or science fiction literature. However, it has only really become accessible to the general public in recent years, with the arrival of generative models.

Generative AI, or “GenAI”, is the result of several decades of progress in artificial intelligence. We started with classic AI, based on procedural rules, and then we evolved towards machine learning, where algorithms learn through a system of rewards and penalties based on the results. Then, “deep learning” made it possible to simulate the functioning of human neurons, thanks to considerable computing power. Today, generative AI combines all these technologies and offers a more accessible user interface, allowing anyone to ask questions in natural language and get answers without having any particular technical skills.

The change isn’t in the output but in the input. Previously, users had to follow precise methodologies; today, anyone can ask questions, even ambiguous ones, through simple prompts, without requiring technical skills. This accessibility has empowered individuals, enabling what is often referred to as “human augmentation.” However, this rapid evolution also introduces challenges and risks.
Generative AI lacks common sense—it cannot discern good from bad, true from false, or public from private information. It generates responses based solely on the input it receives, which means ambiguous or poorly formulated queries can produce misleading results. These responses, though convincing, may amplify errors.
Studies by the Big Four reveal that generative AI enhances performance for high-performing teams but exacerbates inefficiencies in underperforming ones, widening the gap rather than leveling the field. Human expertise remains essential for applying common sense and ensuring responsible use.
Generative AI is not a magical solution to all problems. Its effectiveness depends on how it is used. As the initial hype diminishes, maturity and experience in leveraging these tools will lead to better outcomes.

Nevertheless, “GenAI”, due to its accessibility and capabilities, presents many risks. For example, models available to the general public (such as ChatGPT) are very sensitive to what is called “hallucinations”, i.e. the generation of an answer that seems in every way correct and precise, but which is, in reality, totally erroneous. The same goes for the preservation of intellectual property: these models are often fed by the information available on the web, but how can we differentiate a result that uses material protected by copyright?

It is therefore essential to know the limits of these tools in order to be able to properly assess their potential and risks. This is particularly true for areas related to Human Resources.

VM: What are the main advantages of generative AI for HR?

GC: Generative AI has enormous potential in the field of HR, thanks to applications allowing the improvement of the user experience, the generation of documents and the automation of tasks. For example, it allows to analyze CVs much more precisely and make immediate recommendations for vacant positions, which is crucial in the current context of the war for talent. It can also speed up the creation of job descriptions and offer natural language chatbots to support employees and candidates. Some tools even allow to analyze job interview sessions, thus providing additional elements to help recruiters make decisions. HR can thus save precious time and focus on the most promising candidates and activities with higher added value.

However, we must also be aware of the risks associated with generative AI, particularly in the HR field where it has a direct impact on candidates and employees. This potential for innovation and power makes AI a tool whose limits and rules are essential to know, not to mention legal regulations. In my point of view, one of the main dangers is that AI does not have common sense. It cannot distinguish between good and bad, true and false; it simply responds based on the data it has or generates. This can lead to erroneous or biased results if the training data is imperfect. For example, a recruiting algorithm trained on the company’s history risks reproducing conscious or unconscious biases of recruiters and managers. Unlike a human, an insufficiently parameterized AI can amplify these biases on a large scale and without limits, making the recruitment process potentially discriminatory.

VM: How can companies avoid these pitfalls?

GC: My first piece of advice is to start with education: understand the basics of AI, its limitations and risks, and not let yourself be influenced by marketing. This education must be done at all levels of the organization, but of course especially at the Leadership level. It is essential to have a good understanding of ethical issues and current regulations, because AI is becoming omnipresent. Every manager must have a minimum of knowledge in this area.

Then, it is crucial to resist the pressure linked to the buzz around AI. Today, everyone wants to position themselves on “GenAI” because it is what sells and attracts visibility. However, an immature AI strategy, deployed on a bad process or with insufficient training and testing, can represent a huge risk for an organization, both in terms of results and costs or image. You have to take a step back, involve the right expertise, and develop a strategy that takes into account the many risks associated with a technology that is still in development.

My third piece of advice, especially in the HR field, is to involve experts to ensure that AI is used ethically and in compliance with regulations, especially those of the European Union. It is important not to rely on the information provided by technology vendors and it is always important to have an independent view. Furthermore, since we are talking about human management, it is essential not to delegate the decision-making process entirely to AI. The results provided by AI must be interpreted and validated by HR professionals to ensure their relevance and fairness.

VM: What are the best practices for integrating generative AI into HR?

GC: The integration of AI into HR must be accompanied by adequate training of teams so that they know how to use these tools effectively. At Danone, for example, we have set up compliance and ethics committees to assess and supervise the use of AI. We are also committed to training our employees on the potential and limitations of AI.

It is crucial to view AI as a tool to augment human capabilities and not as a magic solution. AI can improve productivity and efficiency, but it must be used with discernment and responsibility.

VM: What future do you see for generative AI in HR?

GC: Generative AI represents a technological revolution comparable to the advent of the Internet. It will undoubtedly transform many aspects of HR. However, it is imperative to adopt it thoughtfully and ethically, without fear but with caution. Jobs are evolving, and skills will have to adapt to this new reality. AI should not be seen as a threat, but as a powerful tool at the service of humans and productivity.

In conclusion, it is important to understand that this is a very broad, complicated but also fascinating subject. A successful integration of Generative AI (and even “normal” AI) in HR relies on a combination of technological innovations, ethical rigor, and continuous training. By respecting these principles, we can maximize the benefits of AI while minimizing its risks.

VM: Thank you very much, Gregory, for this detailed insight on Generative AI and its implications in Human Resources.

GC: Thank you, Vincent. It was a pleasure to share these thoughts.