罗马教皇利奥十四世近日再次对人工智能在军事领域的应用发出警告，呼吁各国政府采取具体行动，禁止能够自主决定攻击目标并实施致命打击的武器系统。他强调，涉及人类生命、自由以及武力使用的重大决定，必须始终处于有效、知情且负责任的人类监督之下，不能交由人工智能独立作出。
当地时间10月30日，利奥十四世在梵蒂冈接见了参加人工智能专题会议的科学家、技术专家及相关机构代表。这场名为“人工智能的未来：梵蒂冈与科学界关于智能技术发展及应用的对话”的会议，由梵蒂冈宗座生命科学院与未来生命研究所共同组织，重点讨论人工智能发展所带来的技术、伦理和社会问题。
在讲话中，教皇回顾了前任教皇方济各2024年向七国集团领导人发表的讲话。当时，方济各将人工智能描述为一种既令人兴奋又令人担忧的工具，并明确表示，任何机器都不应被赋予决定是否夺取人类生命的权力。利奥十四世认为，这一警告至今仍然具有现实意义。
利奥十四世指出，人工智能的未来并非早已注定。作为人类开发的技术，其发展方向最终取决于人类自身的选择。因此，人类不仅有责任引导人工智能的发展，也必须在必要时改变甚至拒绝某些应用方式。真正需要讨论的问题，不应仅仅是人工智能能够做到什么，更重要的是人类究竟希望它服务于什么目的。
他特别强调，人工智能必须接受有效的伦理约束，尤其不能让算法系统在缺乏人类实质监督的情况下决定谁应该成为攻击目标，或者谁应当受到边境管控。将一个人列入军事目标名单或边境管制数据库，可能使其被简化为一组数据，而背后涉及的真实人生和个人权利却被忽视。
因此，教皇呼吁各国对人工智能进行所谓的“解除武装”，并特别敦促即将在11月于瑞士日内瓦举行相关会议的国家采取具体措施，推动禁止致命自主武器。
所谓致命自主武器，是指能够在没有人类直接干预的情况下，识别、选择并攻击目标的武器系统。随着人工智能、计算机视觉、无人机和自动化军事系统的发展，这类技术已经成为国际社会讨论军事伦理和战争规则时的重要议题。其核心争议在于，一旦机器获得自主决定攻击目标的能力，责任归属、误判后果以及人类对战争的控制都可能面临新的挑战。
除了军事用途，利奥十四世还对人工智能是否应当获得法律人格表达了明确反对意见。他指出，机器只是工具，而不是人。人工智能并不具备真实的身体、情感、生活经历和人际关系，也无法从自身经验中理解爱、友谊、劳动和责任。
他认为，如果赋予人工智能系统法律人格或独立权利，反而可能模糊真正的责任主体。当人工智能造成损害时，受害者可能不得不面对一个无法承担道德责任的技术系统，而开发者、运营者和实际使用者的责任则可能因此变得更加模糊。
教皇还强调，人工智能的发展必须服务于全社会的共同利益，而不能仅仅成为少数企业和利益集团获取权力与财富的工具。技术进步所带来的医疗、知识、工具和发展机会，应当更加公平地惠及不同国家和社会群体，特别是那些最需要帮助的人。
在他看来，目前全球并不缺少关于人工智能伦理的原则性讨论，真正迫切的问题是建立能够将这些原则转化为实际约束的制度。国际社会需要制定共同标准，建立独立监督机制，并明确人工智能开发和使用过程中的责任追究制度，使相关决策能够接受审查，在造成损害时也能够提供有效补救。
这次讲话延续了利奥十四世今年5月发布首份通谕《Magnifica Humanitas》（《伟大的人性》）中的核心立场。该文件将人工智能作为重点议题之一，呼吁各国加强法律监管，防止技术发展损害人的尊严，并警告人工智能在战争、劳动市场和公共决策中的应用可能带来深远影响。
利奥十四世此次将目光进一步集中到致命自主武器问题上，也使梵蒂冈的立场与国际社会正在进行的军事人工智能治理讨论产生了更直接的联系。联合国《特定常规武器公约》框架下的相关会议将于11月16日至20日在日内瓦举行，致命自主武器的国际监管将是讨论议题之一。
教皇并未要求全面停止人工智能研究，而是主张对可能造成不可逆人身伤害的应用设立明确边界。他希望各国能够通过国际合作建立有效监管体系，确保技术始终处于人的责任和控制之下，而不是让人工智能的发展速度超越人类社会管理风险的能力。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
In recent days, Pope Leo XIV has once again warned against the use of artificial intelligence in the military sphere and called on Governments to take concrete action to ban weapons systems that are capable of autonomous targeting and deadly strikes. He stressed that major decisions involving human life, freedom and the use of force must always be under effective, informed and responsible human oversight and cannot be left to artificial intelligence to make independently.
On October 30, local time, Leo XIV received scientists, technical experts and representatives of relevant institutions attending the conference on artificial intelligence in the Vatican. The conference, entitled “The future of artificial intelligence: a dialogue between the Vatican and the scientific community on the development and application of intelligent technology”, organized by the Pontifical Academy of Life Sciences of the Vatican and the Institute for Future Life, focused on the technical, ethical and social aspects of artificial intelligence development.
In his address, the Pope recalled the statements made by former Pope Francis to the leaders of the Group of Seven in 2024. At that time, Francis described artificial intelligence as an exciting and worrying tool and made it clear that no machine should be empowered to decide whether or not to take human life. According to Leo XIV, this warning remains relevant today.
According to Leo XIV, the future of artificial intelligence is not pre-determined. The direction of the technology developed as a human being ultimately depends on the choices of the human race itself. Thus, human beings have a responsibility not only to direct the development of artificial intelligence, but also to change and even reject certain applications where necessary. What is really needed is not just what artificial intelligence can do, but more importantly what human beings want it to serve.
In particular, he stressed that artificial intelligence must be subject to effective ethical constraints, in particular by not allowing algorithms to determine who should be targeted or who should be subject to border control in the absence of substantive human oversight. The inclusion of a person on a list of military targets or a border control database may simplify it into a data set, while the real life and individual rights underlying it are ignored.
The Pope therefore calls on States to carry out so-called “disarmament” of artificial intelligence and, in particular, urges States that are about to hold a relevant conference in Geneva, Switzerland, in November to take concrete measures to promote the prohibition of deadly autonomous weapons.
Deadly autonomous weapons are weapons systems that can identify, select and attack targets without direct human intervention. With the development of artificial intelligence, computer vision, drones and automated military systems, such technologies have become an important topic for the international community in discussing military ethics and the rules of war. At the core of its controversy is that once the machine acquires the capacity to decide its own targets, the attribution of responsibility, the consequences of miscalculation and human control over war may face new challenges.
In addition to military use, Leo XIV expressed a clear objection to whether artificial intelligence should acquire legal personality. He pointed out that machines were tools, not people. Artificial intelligence does not have real physical, emotional, life experience and interpersonal relationships, nor does it understand love, friendship, labour and responsibility from its own experience.
In his view, if legal personality or independent rights were granted to artificial intelligence systems, it might blur the true subject of responsibility. When artificial intelligence causes damage, victims may have to face a technological system that cannot assume moral responsibility, and the responsibilities of developers, operators and actual users may become more blurred.
The Pope also stressed that the development of artificial intelligence must serve the common good of society as a whole and not merely serve as an instrument of power and wealth for a few enterprises and interest groups. Health care, knowledge, tools and development opportunities brought about by technological advances should benefit more equitably different countries and social groups, especially those most in need.
In his view, there was no lack of global discussion of the principles of artificial intelligence ethics, and the real urgency was to establish systems that would translate those principles into practical constraints. The international community needs to develop common standards, establish independent oversight mechanisms and clarify accountability systems in the development and use of artificial intelligence so that relevant decisions can be reviewed and effective remedies can be provided in case of damage.
This statement is a continuation of the central position of Leo XIV in his first message, Magnifica Humanitas, in May of this year. The paper, which identified artificial intelligence as one of its main themes, called on States to strengthen legal regulation to prevent technological developments from undermining human dignity and warned of the potentially far-reaching implications of its use in war, labour markets and public decision-making.
This time, Leo XIV has further focused his attention on the issue of lethal autonomous weapons, which has also led to a more direct link between the Vatican ' s position and the international community ' s ongoing discussions on military artificial intelligence governance. Relevant meetings within the framework of the United Nations Convention on Certain Conventional Weapons will be held in Geneva from 16 to 20 November, and international regulation of lethal autonomous weapons will be one of the topics for discussion.
The Pope did not call for the complete cessation of artificial intelligence research, but rather for the establishment of clear boundaries for applications that could cause irreversible bodily harm. He hoped that States, through international cooperation, would establish effective regulatory systems to ensure that technology remained under human responsibility and control, rather than allowing artificial intelligence to develop faster than human society ' s ability to manage risks.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.