Publications
FOCUS 73: Augmented Intelligence: A Hybrid Approach To Traditional Modeling
Paul Labrogère, CEO of the Research and Technology Institute SystemX, discusses the challenges of its industry partners in deploying AI.
The SystemX Research and Technology Institute is a major player in France in applied research which collaborates and supports the digital transportation of industry partners ranging from SMEs to large international groups. What are the challenges for deploying artificial intelligence technologies?
AI is a national and international priority as recently demonstrated with the setup of an observatory on AI by the OECD in Paris. This international initiative comes from the G20, of which Singapore is one of the founding members. AI is a real asset to meet the challenges of digital transformation, by allowing them to optimize the design, the management of complex systems and their decision-making. We observe that AI mainly benefits companies that develop "consumer" products and services. However, AI is not spreading quickly enough in industries with a strong engineering component.
Their industrial applications implement tooled processes to model the behavior of complex systems from their design to their operation. These tools are generally based on a single "culture" resulting either from scientific computing (physical modeling of systems), or from the representation of knowledge (business knowledge or models of behavior expressed by experts), or from machine learning (data-based models). The AI revolution must go through Augmented Intelligence which consists in implementing approaches hybridizing these different "cultures" of modeling. It requires a global transformation of the company with its engineering teams.
How do you collaborate with the industry on hybridizing?
We must first understand the uniqueness of the model of Research and Technology Institutes (IRT) set up in 2012 by France. These institutes aim to bring together actors from the industrial, academic or institutional worlds to collaborate on research subjects applied to real use cases. SystemX, is dedicated to the digital transformation of industries, services and our nations. We provide technological bricks and scientific skills in fields such as AI but also big data, blockchain, cybersecurity, digital twins. For example, we have been in Singapore since 2017 where we partner with the French Industry (Renault, SNCF, Systra) and CETRAN at NTU on the validation by simulations of the autonomous vehicle.
It is within this unique model that we launched a vast program entitled "Artificial Intelligence and Augmented Engineering" (IA2) which aims to shake up traditional methods by developing solutions to use cases hybridizing the three traditional approaches to modeling: data-based modeling, expert modeling, physical modeling.
Any illustration of use case?
Many of our projects use artificial intelligence. I can cite for example our work in the field of electromobility. We have used IA modeling along with other modeling technics to address a use case on the deployment of charging stations and had to estimate the impact of the use of a fleet of electric vehicles on energy systems. In the field of mobility, we have designed dynamic and predictive models for the simulation of the mobility of people at the scale of a neighborhood, with a focus on the first and last miles. This work aimed at a single objective to improve the operation of multimodal networks. And we expect this work to be reused for a use case dedicated to last mile delivery services which is a growing issue.
Is trust an issue with regards to acceptance of AI by the industry and its customers?
While it is safe to say that Augmented Intelligence will completely revolutionize the industry, a major challenge remains to be taken up: ensuring the transparency and explainability of artificial intelligence-based systems. Understanding, explaining and evaluating how they work is essential. Assurance of safe and reliable behavior is central to complex systems, even more so when they are critical. Trust must therefore be at the heart of the design of systems incorporating AI. Only then, they can be widely adopted. This critical challenge is another priority of our research, in collaboration with academic and industrial partners, for the months and years to come.
Interview with Paul Labrogère, CEO of the Research and Technology Institute SystemX, for FOCUS #73. To read more articles from this issue, download your digital copy here