The debate over the use of autonomous weapons is raging in many countries, and it is clear that they pose serious security and ethical issues. China, for example, has proposed more stringent criteria such as a threshold level autonomy. Other countries are considering more conservative definitions, such as a lethality threshold or evolution requirements.
Arguments for a ban
There are strong arguments for and against an automatic weapons system ban. Artificial intelligence technology is rapidly improving, and scientists are worried about its possible application. The United States has so far resisted the calls for an end to the use of these weapons. However, over 30 other countries have opposed the use of such weapons. New Zealand's arms control Minister stated recently that the use of these weapons is against its values. UN Secretary General has also called for an end to such weapons.
While the private sector is an important part of the debate, it has expressed concern over the proliferation unregulated autonomous weapons. For an international treaty to ban these weapons, the Future of Life Institute began gathering signatures in this summer.
Development challenges
A number of issues will have to be dealt with as autonomous weapons get more sophisticated and more common. Software failure is one of the biggest issues. These weapons are more complicated than traditional human-guided weapons. Software failures can lead to critical mistakes or misinterpretation. This could have catastrophic consequences.
Human error can also pose a challenge. A number of legal limitations that govern autonomous weapon design may force the use of conservative configuration operational limits. This could affect the weapon's functionality. Additionally, autonomous subsystems could be "undeclared consumers" which would allow them to consume the outputs of their prediction process as input. Unintended feedback loops may result, similar to filter bubbles on social networks. Commanders might not be able to correct weapon behaviour in such an environment.
Security concerns
There are security concerns with autonomous weapons. These weapons have the potential to fall into the wrong hands and become deadly. Because of the structure of Western militaries, it is unlikely they will use autonomous weapons. Even if the technology was developed in future, it is possible that such weapons could be misused.
These weapons could be used against civilians, which is why many countries are concerned. Iraq, for example, has warned against the use of fully autonomous weapons, saying such weapons could trigger an arms race and have catastrophic effects. Iraq has stated that no decision can ever be handed to machines and that decision-making must always remain human-centered. In November 2017, Iraq called for a pre-emptive ban on lethal autonomous weapons systems, and has spoken out against them in other forums. Although Iraq participated in the UN Security Council meeting about autonomous weapons in August, the country has not yet joined formal the resolution.
Ethics
Currently, the ethical considerations of autonomous weapons are controversial. Some argue that autonomous weapons are immoral. Others say they are both moral and rational. This is a complex issue, and it is important that you consider all ethical implications before moving forward with technological development. This paper discusses the dualistic notions of moral responsibility. This paper emphasizes that not all moral responsibility entails a duty to loyally serve a legitimate authority. It examines the shifting dynamics of autonomy, accountability and warfare in 21st-century warfare.
In conflict situations, it is difficult to develop autonomous weapons. In particular, autonomous weapons will likely accelerate hostilities' onset, which will place the burden of war on civilians. Additionally, AI systems will make mistakes and increase tensions. Mass killing is a context in which ethics can be challenged. AI systems should not be considered an exception. These weapons shouldn't be created without human oversight.