The Politics and DV Engineering working group analyses current political developments in the defence sector. It provides independent facts and interpretation from the technological knowledge and experience of engineers.
The Politics and DV engineering working group has questions and comments on D66's initiative note on autonomous weapons of 18 May 2021 and the government' s response of 28 June 2021 to this note..The commentary has been prepared based on publicly available information and defence technology knowledge and experience.
Download a PDF of this commentary here.
The paper seems to be written mainly from the perspective of armed UAVs, as an armed MQ-1 Predator might be. However, autonomous weapons have been available in the naval domain for over 30 years, and in the land domain the possibilities will grow rapidly. Think Goalkeeper, sea and land mines, armed unmanned vehicles (USVs), etc. Especially on land, humanitarian risks are higher and technical manageability more limited.
The working group endorses the government's view, expressed in the response to the paper, that it is important to build knowledge about autonomous weapons. So does the government's view that reference to international law, in particular the humanitarian law of war, is sufficient.
However, the way in which both aspects are given substance needs to be specified.
The statement made by the government under decision point 1 indicated discussions on legal concepts and frameworks only contribute to solutions to a limited extent. The working group is of the opinion that a stronger and more active stance is needed. After all, technology does develop. Especially among potential adversaries. It is necessary for engineers, military personnel and the organisations in which they operate to have a foothold and to be able to invoke these regulations or agreements when necessary.
At decision point 2 the working group points out that the requested procedures and technology to embed human control in systems are generally easy to circumvent. (cf. tamper-evident). Even access to source code no longer makes much sense with self-learning systems. We think it would be advisable to have an independent multinational organisation, such as the United Nations, set standards and conduct the analysis referred to in Article 36 of Additional Protocol I to the Geneva Conventions. Not the developer of the product, as stated in that article.
Regardingdecision point 3 is the government's position on programming the characteristics of the targets to be attacked and in deciding on the deployment of the weapon. If defence against a concrete attack always requires an emphatic decision prior to each deployment, this may pose a survival risk. After all, humans already act slower than technology and will make slower and less accurate decisions than machines when faced with a high threat. Proper programming of targets to be attacked and a veto button to cancel deployment seems more effective. In addition, it is important to properly define the circumstances of threat under which a weapon system may be put into fully automatic mode, as has been the case with the Goalkeeper on board naval ships since 1980.
The amendment proposed by D66 in decision point 4desired restriction on development of fully autonomous weapons will hamper the accumulation of the necessary knowledge. Even if it is known that potential opponents do engage in such development. Lack of knowledge poses a risk to any necessary defence against an opponent's fully autonomous weapons.
With regard to decision point 5 the working group doubts the government's assertion that all systems acquired by the Netherlands have a registration functionality that sends information on use to a centralised database. Perhaps the government could specify this position and indicate which weapon systems send information to which databases (F-35, MQ-1, naval vessels?) and by whom they are managed. Moreover, the intended goal of being able to identify violations of international law seems to the working group difficult or impossible to achieve from a technical point of view. Especially for ground-based applications and if the database is placed remotely. Such a registration requires transport large amounts of high-resolution data, for which the necessary connections are as yet lacking. Moreover, to avoid detection in gravity conditions, it is usually undesirable to establish active (radio) links.
_____________________________
The Hague, 23 August 2021
More information on the Politics and Defence Technology working group can be found via this link.
Do you have any questions? If so, please contact the working group at E: politiektechniek@kividv.nl
Would you like to receive the working group's comments and opinions by email? Sign up via this link.
Disclaimer: The facts and opinions given are based on open sources and on the knowledge and experience of working group members.
As part of the professional association KIVI, the working group is independent of political parties, governments and companies.
This is not an official position of KIVI. The association accepts no liability for anything put forward by the working group or its members.
Photo: Ministry of Defence
