Free beta access. 5 free AI tokens each week. Optional top-ups available. See what is included
Opening Word document…

LAWS meet only one part of the IHL threshold – do not meet on targeting law – their evidence

Ettinger, 20 (Jay Ettinger, 2020, accessed on 6-29-2022, 30 Minn. J. Int'L. 153, "1. NOTE: Overcoming International Inertia: The Creation of War Manual for Lethal Autonomous WeaponsSystems, 30 Minn. J. Int'l L. 153", file:///C:/Users/ncb12/Downloads/Nathan%20Boyle%20Overcoming%20International%20Inertia_%20The%20Creation%20of%20War%20Manual%20for%20Lethal%20Autonomous%20Weapons%20Sy.PDF) (NB)

In order for LAWS to be compliant with IHL "targeting law," "they must be able to reliably and predictably distinguish between combatants and non-combatants, as well as make rapid judgments on the proportionality of an attack against its potential collateral harms. 67First, there is a question as to whether computer algorithms will be able to gauge the complex, context-dependent, and humanistic clues that soldiers must use to distinguish combatants and non-combatants in the modern battlefield where combatants often attempt to conceal their identities. 68Second, even if such a distinction is technically [*163] feasible, there is a question as to whether these systems can reliably make sound decisions given the vast array and often rapidly changing nature of battlefield contexts. 69For example, one potential risk to the system's reliability is the introduction of bias to decision-making originating in the data sets used to train the AI system. 70Even if both of these technical feasibility questions can be adequately addressed, there is a further question as to whether lethal decision-making inherently requires human involvement under IHL. 71And if there is such a requirement, what degree of human involvement is sufficient to meet IHL obligations must also be determined.

Sign in to recut, update, or save this card.