اصول هوش مصنوعی آسیلومار

The Asilomar AI Principles

Artificial intelligence has already provided beneficial tools that are used every day by people around the world. Its continued development, guided by the following principles, will offer amazing opportunities to help and empower people in the decades and centuries ahead.

RESEARCH ISSUES

§1Research Goal: The goal of AI research should be to create not undirected intelligence, but beneficial intelligence.

§2Research Funding: Investments in AI should be accompanied by funding for research on ensuring its beneficial use, including thorny questions in computer science, economics, law, ethics, and social studies, such as: (a)How can we make future AI systems highly robust, so that they do what we want without malfunctioning or getting hacked?

(b)How can we grow our prosperity through automation while maintaining people’s resources and purpose?

(c)How can we update our legal systems to be more fair and efficient, to keep pace with AI, and to manage the risks associated with AI?

(d)What set of values should AI be aligned with, and what legal and ethical status should it have?

§3Science-Policy Link: There should be constructive and healthy exchange between AI researchers and policy-makers.

§4Research Culture: A culture of cooperation, trust, and transparency should be fostered among researchers and developers of AI.

§5Race Avoidance: Teams developing AI systems should actively cooperate to avoid corner-cutting on safety standards.

ETHICS AND VALUES

§6Safety: AI systems should be safe and secure throughout their operational lifetime, and verifiably so where applicable and feasible.

§7Failure Transparency: If an AI system causes harm, it should be possible to ascertain why.

§8Judicial Transparency: Any involvement by an autonomous system in judicial decision-making should provide a satisfactory explanation auditable by a competent human authority.

§9Responsibility: Designers and builders of advanced AI systems are stakeholders in the moral implications of their use, misuse, and actions, with a responsibility and opportunity to shape those implications.

§10Value Alignment: Highly autonomous AI systems should be designed so that their goals and behaviors can be assured to align with human values throughout their operation.

§11Human Values: AI systems should be designed and operated so as to be compatible with ideals of human dignity, rights, freedoms, and cultural diversity.

§12Personal Privacy: People should have the right to access, manage, and control the data they generate, given AI systems’ power to analyze and utilize that data.

§13Liberty and Privacy: The application of AI to personal data must not unreasonably curtail people’s real or perceived liberty.

§14Shared Benefit: AI technologies should benefit and empower as many people as possible.

§15Shared Prosperity: The economic prosperity created by AI should be shared broadly, to benefit all of humanity.

§16Human Control: Humans should choose how and whether to delegate decisions to AI systems, to accomplish human-chosen objectives.

§17Non-subversion: The power conferred by control of highly advanced AI systems should respect and improve, rather than subvert, the social and civic processes on which the health of society depends.

§18AI Arms Race: An arms race in lethal autonomous weapons should be avoided.

LONGER-TERM ISSUES

§19Capability Caution: There being no consensus, we should avoid strong assumptions regarding upper limits on future AI capabilities.

§20Importance: Advanced AI could represent a profound change in the history of life on Earth, and should be planned for and managed with commensurate care and resources.

§21Risks: Risks posed by AI systems, especially catastrophic or existential risks, must be subject to planning and mitigation efforts commensurate with their expected impact.

§22Recursive Self-Improvement: AI systems designed to recursively self-improve or self-replicate in a manner that could lead to rapidly increasing quality or quantity must be subject to strict safety and control measures.

§23Common Good: Superintelligence should only be developed in the service of widely shared ethical ideals, and for the benefit of all humanity rather than one state or organization.

The signature list grew dramatically after we posted the principles online, and by now it includes an amazing list of more than a thousand AI researchers and many other top thinkers. If you too want to join as a signatory, you can do so here: http://futureoflife.org/ai-principles.

We were struck not only by the level of consensus about the principles, but also by how strong they were. Sure, some of them sound about as controversial as “Peace, love and motherhood are valuable” at first glance. But many of them have real teeth, as is most easily seen by formulating negations of them. For example, “Superintelligence is impossible!” violates §19, and “It’s a total waste to do research on reducing existential risk from AI!” violates §21.

Indeed, as you can see for yourself if you watch our long-term panel discussion on YouTube,11 Elon Musk, Stuart Russell, Ray Kurzweil, Demis Hassabis, Sam Harris, Nick Bostrom, David Chalmers, Bart Selman and Jaan Tallinn all agreed that superintelligence would probably be developed and that safety research was important.