Too often, companies focus on defining a global artificial intelligence (AI) strategy that can be applied to all their operations in “Swiss Army Knife” mode. In practice, this approach rarely leads to concrete results.
According to the IT consultant Gartner, when a company decides to integrate artificial intelligence into its activities, it must favor an iterative approach based on 5 main stages.
The first is to identify specific business projects whose scope is clearly defined and whose potential impact is significant to the business. These projects must make it possible to generate measurable and impactful results, in particular through specific indicators whose development can be monitored. A classic example could be optimizing inventory management.
A dedicated team
The second step is to create a dedicated team that brings together the talent needed to complete the projects. This team will have to combine profiles that master AI technologies (machine learning, natural language processing systems etc.), the company’s IT infrastructure and the business requirements linked to the projects. Depending on the size of the company, some of these skills need to be outsourced, especially at the AI level.
The third step is to identify, capture and manage the data needed for the selected projects. The quality and relevance of the data must take precedence over the quantity. In fact, AI doesn’t necessarily rhyme big databut always with smart data. Of course, this data must meet quality standards to ensure that it “represents” completely and correctly the context of the intended projects. Depending on the AI technologies implemented, the minimum amount of data required may vary.
This leads to the fourth step, which should make it possible to identify the AI technologies adapted to the specific objectives of the projects chosen by the company. For example, probabilistic reasoning techniques will be particularly suitable for bringing out “hidden” patterns in a large amount of data, such as fraud patterns. On the other hand, perfecting routes within a supply chain problem will instead require the use of optimization techniques.
Finally, the fifth step should allow the company to structure and pass on the expertise acquired during the implementation of these first projects in order to more quickly implement it for other goals. This step should also make it possible to identify problems or gaps in terms of skills, data and technologies, but also in relation to the general culture of the company in this specific discipline, which is AI.
This 5-step strategy is at the heart of the calls for projects proposed by the DigitalWallonia4.ai program whose ambition is to accelerate the adoption of artificial intelligence by companies and organizations and to develop a reference ecosystem in Wallonia. These calls for projects (Start IA, Tremplin IA, Cap IA) aim to provide concrete support to companies that want to integrate artificial intelligence in their business up to the development of operational prototypes. They allow companies to work with technology partners and research centers to take advantage of cutting-edge AI skills.
The HEC Digital Lab is also part of this dynamic, especially through its initiative data science whose goal is in particular to join initiatives in data science and to promote notable projects in this field.