France has released an AI strategy which supports human responsibility for action
Stanley-Lockman, 2021 - Center for Security and Emerging Technology [Zoe CSET Issue Brief August “Responsible and Ethical Military AI Allies and Allied Perspectives” https://cset.georgetown.edu/publication/responsible-and-ethical-military-ai/ Acc. 4/21/22 TA]
The French Ministry of Armed Forces has been studying AI associated risks since 2019.37 That September, France became the first (and, still at the time of writing, only) European ally to publicly issue a dedicated military AI strategy.38 Ethics and responsibility appear throughout the strategy, most notably as aspects of “controlled AI” and in the announcement to establish a ministerial Defence Ethics Committee.39 Subsequently, as described below, the advisory opinions of the Defence Ethics Committee also lend insights into French views on responsibility and other concepts. Each of these building blocks is indicative of French thought and implementation pathways for ethical and responsible military AI, even if not in the form of adopted principles. The French military AI strategy, called Artificial Intelligence in Support of Defence, describes “controlled AI” as the overarching framework for Ministry guidelines on AI adoption, including aspects that relate to ethics.40 Notably, control (maîtrise) in this context refers to harnessing and governing AI, and is not synonymous with human control (see Appendix II).41 In addition to adoption priorities like the imperative to maintain freedom of action and interoperability with allies, the guidelines see “trustworthy, controlled, and responsible AI” as interlinked concepts under the headline “guidelines for controlled defence AI.” These three concepts come to light in the need for the Ministry to “have robust and secure systems which can be trusted to assist service personnel and commanders, dispelling any ‘black-box’ effect, while retaining human responsibility for action.”42 The rest of the section unpacks what trustworthiness, control, and responsibility mean, as gleaned from both the French military AI strategy and the Defence Ethics Committee advisory opinions on other technologies.43