Quel rôle pour les experts en Science des données dans le nouveau paradigme de l'IA ? : Table ronde

Quel rôle pour les experts en Science des données dans le nouveau paradigme de l'IA ? : Table ronde

Title history
  • Quel rôle pour les experts en Science des données dans le nouveau paradigme de l'IA ? : Table ronde — observed on 20261110-0712 UTC (YouTube)

by ENSAI · View the channel on PSI

Humanities, Social Sciences & Thought AI & Society UBDigital and Information technology: general topicsUBJDigital and information technologies: social and ethical aspects
🎙 ENSAI 👥 1K 📅 November 6, 2026 ⏱ 56 min 👁 125 📄 debate 🧭 2026-11-10
Available in: English (current) Français

Keywords

explicabilityAI governanceagentic AIhuman in the loopdata science roles

Summary

This roundtable discussion, held during the 30th anniversary of ENSAI, brings together experts from academia, industry, and defense to examine the evolving role of data science professionals in the age of generative and agentic AI. The panelists address how explainability is being redefined, moving from model internals to systemic transparency through techniques like RAG and reasoning chain analysis. They discuss the trade-off between performance and interpretability, emphasizing that regulatory pressures, such as the EU AI Act, increasingly demand clarity, especially in high-stakes sectors like finance, water treatment, and defense. Governance emerges as a central theme, with attention to responsibility, ethical charters, and the human-in-the-loop principle. The speakers highlight the need for data scientists to adapt by embracing multi-criteria optimization (including sovereignty and sobriety) and by engaging business users early in the design process. The defense perspective notes internalization of AI capabilities to ensure control and trust. The discussion concludes that while challenges are growing, the role of data science experts is becoming more critical in bridging technical capabilities with societal expectations.

171 words

Critical Evaluation

Value of the Information & Strength of the Argument

The value of this roundtable lies in its diverse perspectives: an academic (Elisa Fromont), a bank CDO (Maxime Aves), a consulting CEO (Pascal Bizari), a defense agency representative (Loïc Kouenek), and a utility AI leader (Claire). Their arguments are grounded in concrete examples: credit scoring, fraud detection, water treatment optimization, and military decision-making. The panel effectively demonstrates that explainability is not a binary concept but a spectrum depending on use case and risk. The argumentation is coherent and non-dogmatic, acknowledging the necessary trade-offs and the practical difficulties in implementing transparent AI systems. The discussion also highlights a shift from model-level explainability to process-level (agentic) transparency, reflecting current technological trends. The panelists agree on the increasing importance of governance and ethics, though they admit that many companies are still in a reflective phase rather than fully operationalizing these principles.

Scientific Rigor, Source Quality, Title Accuracy

The rigor is high given the expertise of the participants, but the discussion is conversational and relies on experience rather than cited literature. No external sources are referenced beyond the school’s social media links in the description. The title is well matched to the content, which genuinely addresses the role of data science experts in the new AI paradigm. The lack of formal citations is typical for such roundtables, but the practical insights are valuable. The discussion avoids generalized claims and always ties arguments to specific domains, which reinforces its credibility. The content is direct and avoids sensationalism, focusing on real-world constraints and opportunities.

257 words

Title / Content Match

The title accurately reflects the content: a roundtable exploring the evolving role of data science experts in the context of new AI paradigms, with a focus on challenges and opportunities.

Quality & Reliability

8/10

Panel of recognized experts from academia, industry, and defense provide nuanced, well-informed perspectives on explainability, governance, and ethics in AI. The discussion is balanced and grounded in practical experience, though it remains a roundtable and not peer-reviewed.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

  • ENSAI LinkedIn — Official LinkedIn profile of ENSAI, referenced in the video description.

Contribution & Novelties

The roundtable provides a unique synthesis of perspectives on how data science experts must adapt to the new AI paradigm, highlighting the evolution from model-centric explainability to systemic and process-oriented transparency. It offers concrete examples from regulated industries and defense, illustrating the practical implementation of governance frameworks and human-in-the-loop principles.

Pour aller plus loin :

  • Explainable artificial intelligence — This concept underlies the entire discussion; Wikipedia provides a good overview of methods and trade-offs.
  • Artificial Intelligence Act — The EU regulation is frequently mentioned by panelists as a driver for explicability and governance.
  • Human-in-the-loop — A key principle discussed in the roundtable, particularly regarding accountability and defense applications.

108 words

Radar Profile

The radar profile shows high scores in information quality and reliability, with moderate technical depth and quantity, indicating a well-balanced expert discussion accessible to a broad audience.

Reliability 8/10