How to responsibly use AI in the military domain: A Review of the "Ethics and Technology" session of the AFCEA Bonn:

Disclaimer: The following is a summary I wrote on Linkedin, but I think the topic is too important to be "hidden" there, so I repost it here.

This AFCEA Bonn event had the title "Technical dimensions of responsible AI usage in defence". It took place on February 20th at the Wissenschaftszentrum Bonn. The main speakers were retired General Dr. Ansgar Rieks and Professor Dr. Wolfgang Koch. The audience consisted of scientists, soldiers, members of BWI and industry representatives. Both speakers first presented their individual statements before opening the floor to questions from the audience and engaging in a lively debate.

Professor Koch started off the event with a holistic philosophical review of relevant military statements from the perspective of the German armed forces post WW2. As such, especially Wolf Graf von Baudissin and his visionary statements about the impact of modern digital technologies on the ethical conduct of soldiers played a role. He then expanded the focus to include the OODA-loop by Colonel Boyd to highlight the potential of AI to provide volitive, cognitive, normative and reflective assistance to human decision-makers. He ended his statement with the introduction of the Ethical AI Demonstrator, an ongoing project included in the development of the FCAS.

Next up, General Rieks gave an overview about his work on peace ethics under the impression of the worsening security situation of the last years. The effect of Russian aggressions, hybrid warfare and the rise of transactional politics and ethics challenge us and our value-based system massively. And as he explained, this is not limited to land or naval warfare, space and cyber space are also highly contested areas, where the impact of AI cannot be overstated. Based on this assessments, he provided a number of open-ended questions concerning the need for a specific kind of data- or algorithm-ethics. He ended his presentation with a personal proposition of 10 ethical axioms or rules for the ethical usage of AI.

Both speakers were united in their assessment of the necessity and inevitability of military AI. They also called for a quick adoption and development of the necessary technologies to defend ourselves against the looming threat of drone swarms and cyber attacks. Yet they also called for inclusion of ethical values in the form of structured, operationalized values and auditable measures.

How this could be undertaken, what values to account for, how to resolve conflicts between values, what decisions should be outsourced to algorithms, what forms meaningful human control could take - and many other relevant issues were discussed with the engaged audience.

My personal takeaways:

🛫 FCAS is not only an essential technological development, but also one of the potential lighthouses for championing ethical considerations in defense. I hope it gets the attention and support it needs to pave the way for further investments in this field.

⚖️ Both speakers agreed on the need to consider ethics additionally to legal and technical limitations. For me, the way it was described reminded me of the "soft ethics" approach, which I also hold for highly relevant. 

🚨The dilemma of conflating ethics and legal regulation continues, one my biggest woes. This issue is the reason many RfPs ask for ethical considerations but describe the necessary roles as attorneys or software engineers, which usually do not have the necessary skillset. Also, most organizations are willing to pay for legal advice, but not for ethical advice. They seem to be content with having a number of principles at the top, a random number of random tools at the bottom and a yawning void in between. 

In sum, the event was inspiring and thought-provoking. It encouraged me to continue my own work. That in itself was worth the trip. But it also showcased the relevant research still being done in military digital ethics, which gives me some hope. I am looking forward to the contribution of General Rieks and Professor Koch on upcoming BDSV - Bundesverband der Deutschen Sicherheits- und Verteidigungsindustrie papers. As I am co-leading a new working group on AI with Stefan Mueck , there might be some grounds for collaboration as well.