Xavier Horem (Carmignac): "Data is the new Microsoft Office package"

Democratize the use of data. This is a strategic priority for many companies, and in particular for Auchan and Carrefour in the retail world. ”This issue of the democratization of Data is truly a subject of change for us,” stressed Samir Amellal last month at the Big Data fair. “All professions are destined to transform” and AI is a massive transformation vector, added the CDO and CIO of Auchan Retail.

Xavier Horem, consultant and former member of the Big Data & AI department of Allianz France, also insists on the human aspect of the transformation process through data and artificial intelligence.

Never without my HR… and their budget

On the occasion of a conference co-organized on October 8 by DataScientest and Dataiku, the expert highlighted the importance of the HR and training issue. And this at least for a simple reason: “It is often human resources that carry the transformation budgets”” justifies Xavier Horem.

The Chief Data Officer, but not only him, therefore has an obvious interest in convincing and “getting HR on board” if he wants to land his Data strategy. Failing that, he could deprive himself of the “adequate means to carry out this transformation.”

However, HR is not only legitimate because it has a significant budget. They also have a direct role to play, in particular in order to internalize Data and AI skills.

Xavier Horem does not prohibit the use of consultants, useful to “initiate”. However, it is indeed the internalization of skills that will give results, he judges, prescribing a ratio of one third of external to two thirds of internal.

HR intervenes to help identify the profiles to be internalized and their skills. As technology transforms professions, they are also necessary in the management of the employability of impacted employees while “accompanying and retaining talents”.

“Training, accompanying, transforming by developing skills internally makes it possible to link the company’s culture, its profession therefore, to the need to acquire and develop a data and artificial intelligence culture”” declares Xavier Horem.

Data, new Office Pack

“Data is the Microsoft Office pack of the 90s”” But it is still necessary to train employees, and in a targeted way. For the finance specialist, the starting point remains the use cases. The skills acquired through recruitment or training meet a direct need.

In insurance, subscription fraud and compensation fraud are obvious use cases to create membership. This also presupposes the involvement, from the outset, of managers and directors, and not only on application cases. Their involvement also concerns the selection of candidates for training on the basis of objective criteria and at the service of the transformation plan. The managers themselves will have to be accompanied.

“Managing a new skill and a new way of working is a real topic. Managers must be trained for this, perhaps not via very advanced training. However, they will have to understand the fundamentals of data and projects.”

Training and HR therefore appear key in the transformation process through the democratization of Data and AI. Dataiku also subscribes to this objective. The publisher of a Data Science solution aims to support the democratization movement by reaching a more diverse audience of users.

“4 years ago, when I joined Dataiku, we had a vision of training, of enablement, quite basic, very technical, and quite limited to specialized users of data and AI”” confides Pierre Carrere, Director Alliance & Partnership.

Acting on culture and adoption

From discussions with the CDOs of large clients, but also HR, the observation was born that the tool was not enough. Other actions were necessary to act on culture and adoption. For the users of its solution, after the small cohorts of experts, a new phase is being prepared involving the use by “hundreds, or even thousands of users in the coming years.”

For Pierre Carrere, it is impossible to do alone. Dataiku has therefore signed a partnership on training with a specialist in the sector, data science and “agnostic vis-à-vis tools”: DataScientest. In this context, the company has developed a new 17-hour Data Literacy training course composed of three main modules: Data acculturation, Dataiku for Dataprep and ML & AutoML.

During this training, learners will also be required to deal with use cases that correspond to them. Thanks to this association with DataScientest, Dataiku therefore aims to accentuate the diversification of its users’ profiles and thus to keep pace with democratization. This also applies to generative AI, which will be taken into account in its technology via the implementation of new functions during October.

“In banking and insurance, there are many new populations to welcome. We are tackling a mountain. And the phenomenon remains recent, less than a year”, observes Pierre Carrere – who wants to avoid unused licenses. To this end, it promotes an adoption plan, an HR plan and an animation plan “adapted to the ambition.”

Training: generalist or punch?

But in terms of training, the trajectory to follow is neither unique nor rectilinear, warns Yoël Tordjman, the CEO of DataScientest. Different approaches are possible. On the other hand, the involvement of HR remains key, he confirms. Acculturation and awareness-raising of the HR populations will contribute to this. The approach will also focus on the professions, in particular through adapted use cases.

For example, the training actor has carried out large-scale projects involving, among other things, the creation of a skills repository. “It’s quite difficult”” however, he notes. To mitigate it, it is recommended to focus on “a few business lines or a particular direction.”

This massive construction site can be replaced by an approach based on ”surgical“ and “specific” actions. Example with the topic of the moment: generative AI. DataScientest has carried out so-called “punch” training actions, “quite massive, but adapted to different types of professions”.

The gradual increase in skills constitutes another alternative. It comes after the choice of a tool or a technology, for example a cloud migration on AWS. “It consists of taking the time, job by job, to improve your skills”, and rather on the expert profiles.

For Yoël Tordjman, the priority today is “to support companies in their end-to-end data transformation” on the human side. And this goal cannot be achieved only through expert training. DataScientest is thus aligned with the issue of democratization.