Societal Impact Assessment: What Ethical AI Really Changes

Introduction

Societal impact assessment is now ubiquitous in discourse, but still too often fragile in practice. Out-of-the-box indicators, poorly usable data, evaluations produced too late to inform decisions: these limitations weaken the credibility of impact approaches and their real utility for organizations.

Artificial intelligence can help overcome these pitfalls—provided it is used as a methodological rigor tool, not as blind automation.

Revealer of Intangible Wealth (RRI) places ethical AI at the heart of its assessment approaches to enhance the quality of analyses, the traceability of results, and their ability to inform strategic decisions.

Why Impact Assessment Remains a Point of Fragility

1. Dispersed and heterogeneous data: poorly articulated qualitative and quantitative information.
2. Indicators disconnected from action: measures produced without direct link to change mechanisms.
3. Results poorly appropriable: deliverables difficult for management teams to use.
4. Assessment perceived as reporting: external obligation rather than a management tool.

In this context, assessment loses its primary function: to aid in decision-making.

The Real Role of AI in Impact Assessment

AI neither replaces expert analysis nor human judgment. When used correctly, it acts as an accelerator of methodological rigor.

At RRI, AI is mobilized to:

• Structure and explore large qualitative data sets (interviews, verbatims, documentary corpus).
• Enhance methodological coherence between impact hypotheses, collected data, and conclusions.
• Detect weak signals and emerging tensions that are otherwise difficult to perceive.
• Improve the traceability of analyses and choices made throughout the assessment.

AI is thus subordinate to the method: it illuminates, verifies, and connects, without ever substituting for analysis.

Ethical AI by Design, Not by Declaration

Talking about ethical AI is not enough. It is also necessary to define the concrete conditions of use.

The structuring principles adopted by RRI are as follows:

• Transparency of processes: explainable and debatable results.
• Control of biases: AI as a tool for identification, not concealment.
• Data protection: sobriety, security, and regulatory compliance.
• Human responsibility: the final analysis is always assumed by the expert.

This approach allows AI to become a trusted tool, compatible with the requirements of funders and stakeholders.

Concrete Applications in Service of Impact

Interfacia Project

Launched in 2017, Interfacia has allowed for the objectification of the situation of caregiver employees. The challenge was not only to produce data but to make invisible mechanisms readable in order to guide organizational decisions.

AI4Impact

AI4Impact is a national barometer dedicated to the use of AI by actors of public interest. It aims to qualify practices, identify risks, and equip organizations in their technological choices, without a ranking logic.

In both cases, AI is used as a lever for understanding and management, never as an end in itself.

Conclusion

By equipping impact assessment with ethical, rigorous AI that is subordinate to human expertise, RRI enables organizations to make better decisions, secure their trajectories, and sustainably transform their intentions into real impact.