Research Priorities for Robust and Beneficial Artificial Intelligence', AI Magazine (2015) 105-114, http://futureoflife.org/data/documents/research_priorities.pdf. 5 'The changing character of war, fleeting nature of targets, and glut of big data requires the military to integrate machine learning into its targeting process to win wars' (C. Lewis, 'Capturing Flying Insects: A Machine Learning Approach to Targeting', War on the Rocks, 6 Sep 2016, http://waronthe rocks.com/2016/09/capturing-flying-insects-a-machine-learning-approach-to-targeting/). Others have cautioned that increasing automation exacerbates, rather than addresses challenges linked to speed and data load, and have pointed out that the protection of a state's own forces can be enhanced without autonomy in critical functions. See, e.g., United Nations Institute for Disarmament Research (UNIDIR), Framing Discussions on the Weaponization of Increasingly Autonomous Technologies, 2014, pp 5-6, http://www. unidir.ch/files/publications/pdfs/framing-discussions-on-the-weaponization-of-increasingly-autono mous-technologies-en-606.pdf. For a critical appraisal of the claim that AWS would help reduce cost,

Every day, and in a range of contexts, the use of explosive weapons in populated areas harms civilians. Evidence is growing that elevated levels of civilian harm fit a recurrent pattern, suggesting that more coherent and effective humanitarian responses are needed to enhance civilian protection, especially changes in behaviour of users of explosive weapons. This article describes the effects of explosive violence, critically examines how the existing humanitarian law regime tends to address this issue and explores some current developments in building a research and policy agenda to try to reduce civilian harm from the use of explosive weapons.

No abstract

No abstract