2017年1月初舉行的“Beneficial AI”會(huì)議為基礎(chǔ)上建立起來的“阿西洛馬人工智能原則”，名稱來自此次會(huì)議的地點(diǎn)–美國加州的阿西洛馬（Asilomar）市，旨在確保AI為人類利益服務(wù)。本次會(huì)議參加者是業(yè)界最富盛名的領(lǐng)袖，如DeepMind首席執(zhí)行官Demis Hassabis和Facebook AI負(fù)責(zé)人Yann LeCun等。全球2000多人，包括844名人工智能和機(jī)器人領(lǐng)域的專家已聯(lián)合簽署該原則，呼吁全世界的人工智能領(lǐng)域在發(fā)展AI的同時(shí)嚴(yán)格遵守這些原則，共同保障人類未來的利益和安全。
這一系列原則目前共23項(xiàng)，分為三大類，分別為：科研問題（Research Issues）、倫理和價(jià)值（Ethics and values）、更長期的問題（Longer-term Issues）。具體如下：
1) Research Goal: The goal of AI research should be to create not undirected intelligence, but beneficial intelligence.
研究目的：人工智能研究的目標(biāo)，應(yīng)該是創(chuàng)造有益(于人類)而不是不受(人類)控制的智能。
2) Research 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:
研究經(jīng)費(fèi)：投資人工智能應(yīng)該有部份經(jīng)費(fèi)()用于研究如何確保有益地使用人工智能，包括計(jì)算機(jī)科學(xué)、經(jīng)濟(jì)學(xué)、法律、倫理以及社會(huì)研究中的棘手問題，比如：
如何使未來的人工智能系統(tǒng)高度健全(“魯棒性”)，讓系統(tǒng)按我們的要求運(yùn)行，而不會(huì)發(fā)生故障或遭黑客入侵?
如何通過自動(dòng)化提升我們的繁榮程度，同時(shí)維持人類的資源和意志?
如何改進(jìn)法制體系使其更公平和高效，能夠跟得上人工智能的發(fā)展速度，并且能夠控制人工智能帶來的風(fēng)險(xiǎn)?
人工智能應(yīng)該歸屬于什么樣的價(jià)值體系?它該具有何種法律和倫理地位?
3) Science-Policy Link: There should be constructive and healthy exchange between AI researchers and policy-makers.
科學(xué)與政策的聯(lián)系：在人工智能研究者和政策制定者之間應(yīng)該有建設(shè)性的、有益的交流。
4) Research Culture: A culture of cooperation, trust, and transparency should be fostered among researchers and developers of AI.
科研文化：在人工智能研究者和開發(fā)者中應(yīng)該培養(yǎng)一種合作、信任與透明的人文文化。
5) Race Avoidance: Teams developing AI systems should actively cooperate to avoid corner-cutting on safety standards.
避免競爭：人工智能系統(tǒng)開發(fā)團(tuán)隊(duì)之間應(yīng)該積極合作，以避免安全標(biāo)準(zhǔn)上的有機(jī)可乘。
6) Safety: AI systems should be safe and secure throughout their operational lifetime, and verifiably so where applicable and feasible.
安全性：人工智能系統(tǒng)在它們整個(gè)運(yùn)行過程中應(yīng)該是安全和可靠的，而且其可應(yīng)用性的和可行性應(yīng)當(dāng)接受驗(yàn)證。
7) Failure Transparency: If an AI system causes harm, it should be possible to ascertain why.
故障透明性：如果一個(gè)人工智能系統(tǒng)造成了損害，那么造成損害的原因要能被確定。
8) Judicial Transparency: Any involvement by an autonomous system in judicial decision-making should provide a satisfactory explanation auditable by a competent human authority.
司法透明性：任何自動(dòng)系統(tǒng)參與的司法判決都應(yīng)提供令人滿意的司法解釋以被相關(guān)領(lǐng)域的專家接受。
9) Responsibility: 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.
責(zé)任：高級人工智能系統(tǒng)的設(shè)計(jì)者和建造者，是人工智能使用、誤用和行為所產(chǎn)生的道德影響的參與者，有責(zé)任和機(jī)會(huì)去塑造那些道德影響。
10) Value 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.
價(jià)值歸屬：高度自主的人工智能系統(tǒng)的設(shè)計(jì)，應(yīng)該確保它們的目標(biāo)和行為在整個(gè)運(yùn)行中與人類的價(jià)值觀相一致。
11) Human Values: AI systems should be designed and operated so as to be compatible with ideals of human dignity, rights, freedoms, and cultural diversity.
人類價(jià)值觀：人工智能系統(tǒng)應(yīng)該被設(shè)計(jì)和操作，以使其和人類尊嚴(yán)、權(quán)力、自由和文化多樣性的理想相一致。
12) Personal 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.
個(gè)人隱私：在給予人工智能系統(tǒng)以分析和使用數(shù)據(jù)的能力時(shí)，人們應(yīng)該擁有權(quán)力去訪問、管理和控制他們產(chǎn)生的數(shù)據(jù)。
13) Liberty and Privacy: The application of AI to personal data must not unreasonably curtail people’s real or perceived liberty.
自由和隱私：人工智能在個(gè)人數(shù)據(jù)上的應(yīng)用不能充許無理由地剝奪人們真實(shí)的或人們能感受到的自由。
14) Shared Benefit: AI technologies should benefit and empower as many people as possible.
分享利益：人工智能科技應(yīng)該惠及和服務(wù)盡可能多的人。
15) Shared Prosperity: The economic prosperity created by AI should be shared broadly, to benefit all of humanity.
共同繁榮：由人工智能創(chuàng)造的經(jīng)濟(jì)繁榮應(yīng)該被廣泛地分享，惠及全人類。
16) Human Control: Humans should choose how and whether to delegate decisions to AI systems, to accomplish human-chosen objectives.
人類控制：人類應(yīng)該來選擇如何和決定是否讓人工智能系統(tǒng)去完成人類選擇的目標(biāo)。
17) Non-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.
非顛覆：高級人工智能被授予的權(quán)力應(yīng)該尊重和改進(jìn)健康的社會(huì)所依賴的社會(huì)和公民秩序，而不是顛覆。
18) AI Arms Race: An arms race in lethal autonomous weapons should be avoided.
人工智能軍備競賽：致命的自動(dòng)化武器的裝備競賽應(yīng)該被避免。
19) Capability Caution: There being no consensus, we should avoid strong assumptions regarding upper limits on future AI capabilities.
能力警惕：我們應(yīng)該避免關(guān)于未來人工智能能力上限的過高假設(shè)，但這一點(diǎn)還沒有達(dá)成共識(shí)。
20) Importance: 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.
重要性：高級人工智能能夠代表地球生命歷史的一個(gè)深刻變化，人類應(yīng)該有相應(yīng)的關(guān)切和資源來進(jìn)行計(jì)劃和管理。
21) Risks: Risks posed by AI systems, especially catastrophic or existential risks, must be subject to planning and mitigation efforts commensurate with their expected impact.
風(fēng)險(xiǎn)：人工智能系統(tǒng)造成的風(fēng)險(xiǎn)，特別是災(zāi)難性的或有關(guān)人類存亡的風(fēng)險(xiǎn)，必須有針對性地計(jì)劃和努力減輕可預(yù)見的沖擊。
22) Recursive 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.
遞歸的自我提升：被設(shè)計(jì)成可以迅速提升質(zhì)量和數(shù)量的方式進(jìn)行遞歸自我升級或自我復(fù)制人工智能系統(tǒng)，必須受制于嚴(yán)格的安全和控制標(biāo)準(zhǔn)。
23) Common 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.
公共利益：超級智能的開發(fā)是為了服務(wù)廣泛認(rèn)可的倫理觀念，并且是為了全人類的利益而不是一個(gè)國家和組織的利益。
聯(lián)系客服
