Meaningful Human Control
Sensible human control of artificial intelligence-based systems
Sensible Human Control (SHC) emerged in 2016 as an important concept during specialist meetings organized by the UN Convention on Certain Conventional Weapons (CCW). Although the concept has been linked to autonomous weapons, it can be applied more generally to AI-based military systems (both physical and informational) as a crucial requirement for preserving the moral behavior, accountability and effective operational performance envelope of these systems
The main objective is to learn from discussions within several organizations and countries in order to extract a set of practical human-centered guidelines to inform NATO's future actions in this increasingly important field.
Several study themes are addressed, such as organizational, operational and Human-System integration considerations in HCM. Human factors-inspired design guidelines for achieving HCM are proposed, as are systems engineering methods and indicators for validating HCM. Other themes are also addressed, such as the exploitation of MHC by adversaries, complex socio-technical systems, or moral responsibility within teams combining humans and AI.
Among the concepts studied, trust is mainly cited, both between human and machine and between humans within the limits of the organization or system of systems, and linked to this concept are notions of perceived performance, utility, but also approaches in terms of certification, or uncertainty management.
| Department(s) | Partner(s) | Overall amount |
|---|---|---|
| Automatique |
Air Force Research Laboratory, USA |
6 k€
|
| Main support | Rayout | Date(s) |
|
UPHF
|
International |
2020 - 2024
|