MSNBC的Rachel Maddow对她所说的《华盛顿邮报》关于特朗普政府与俄罗斯合作削弱全球规范人工智能驱动武器的努力的“令人毛骨悚然”的报道作出反应。随着人们对这项技术的担忧持续增长，Maddow表示，该报告提供了一种“相当直接的方式，以理解人工智能如何可能开始大规模杀死人类”。
MS NOW’s Rachel Maddow reacted to what she called “hair-raising” reporting from The Washington Post on how Donald Trump’s administration in collaboration with Russia weakened safeguards in a global effort to regulate weapons powered by artificial intelligence As concerns about the technology continue to grow, Maddow said the report laid out “a pretty direct way to understand how AI could start killing humans in large numbers.”
根据《华盛顿邮报》的报道，九月份，数百名外交官在联合国会议上花费数小时，致力于达成一项首创的关于致命自主武器的协议。但在谈判的最后一天，美国和俄罗斯外交官“反复修改文件，删除了多项旨在保障人工智能在武器中使用的条款”，据三位熟悉谈判情况的人士透露，他们以匿名形式讨论了敏感的闭门程序，且《华盛顿邮报》审阅了相关文件。《华盛顿邮报》报道称，美国和俄罗斯团队删除了要求人类在打击前审查人工智能选定的军事目标的条款。
In September hundreds of diplomats spent hours at a United Nations meeting working toward a first-of-its-kind agreement on lethal autonomous weapons, according to the Post. But on the final day of negotiations, U.S. and Russian diplomats “hammered away at the document, removing a range of provisions designed to safeguard the use of artificial intelligence in weapons, according to three people familiar with the negotiations, who spoke on the condition of anonymity to discuss sensitive closed-door proceedings, and documents reviewed by The Washington Post.”The Post reported that U.S. and Russian teams removed a provision requiring humans to review AI-selected military targets before a strike.
马多沃质疑美国为何要“与俄罗斯结成同盟，以确保这是新的道路规则”。她说特朗普“显然只是决定这就是我们在人工智能和军事武器方面应如何设立规则”，并且似乎认为“实现这一愿景的最佳盟友将是他的朋友弗拉基米尔·普京”。
Maddow questioned why the U.S. was “linking arms with Russia to ensure this is the new rules of the road.”She said Trump had “apparently just decided that this is how we should set things up with AI and military weapons” and appeared to believe “our best ally for achieving this vision will be his friend Vladimir Putin.”
MSNBC《今晨》节目主持人还指出特朗普与科技行业的关系。特朗普多次在白宫接待高级科技高管，包括本月初，当时他向六位杰出的科技和人工智能领袖颁发了国家科学奖章。根据《新共和国》的一项分析，这些获奖者共计向特朗普的竞选活动、就职基金以及白宫舞厅捐赠了至少60亿美元，同时也向亲特朗普的超级政治行动委员会以及与总统议程一致的其他倡议捐款。马多沃说特朗普“人工智能亿万富翁们的金钱和奉承构成了一片海洋，他正在这片海洋中漂浮”。
The MS NOW host also pointed to Trump’s ties to the tech industry. Trump has repeatedly hosted high-ranking tech executives at the White House, including earlier this month when he awarded six prominent tech and AI leaders with the National Medal of Science. According to an analysis from The New Republic the awardees collectively donated at least $6 billion to Trump’s campaign, his inauguration fund and his White House ballroom, as well as to pro-Trump super PACs and other initiatives aligned with the president’s agenda. Maddow said Trump “floats on a sea of cash and flattery from AI billionaires.”
该报道发布之际，正如马多沃所说，全国正经历“人工智能引发的一点小恐慌”。该行业的几位顶尖领袖公开呼吁立即放缓发展速度，警告技术进步的速度超过了人类的控制能力。上个月，前Anthropic研究员雅各布·考克斯辞职并警告说：“构建人工智能的人们真诚地相信，它可能在本十年末之前就将我们全人类置于死地。”马多沃说：“有些人认为这一威胁是真实的；有些人则认为这根本不是真实威胁，完全不是这样。”
The report comes amid what Maddow called “a bit of a national freak-out” over AI. Several of the industry’s top leaders have publicly called for an immediate slowdown in development, warning that the technology is advancing faster than humans can control it. Last month, Jacob Coxon a former researcher at Anthropic, resigned from the company and warned that“the people building AI earnestly believe that it could kill us all by the end of the decade.”“Some people believe this threat is real; some people think this is not at all real, that this is not a real threat,” Maddow said.
她接着描述了她认为“很多人”所处的第三种立场——“很难弄清楚这到底是怎么回事”。
She then described a third position that she thinks “a lot” of people are in — “that it’s just kind of hard to figure out what this is about.”
“很难想象这在实际操作中到底会怎么运作，”她说。“到底是怎么让电脑程序杀死我们这些有形的生命体？”
“It’s kind of hard to imagine how that would really work in practical terms,” she said. “How exactly would computer programs kill us physical beings?”
对于这第三类人，马多说，《华盛顿邮报》的报道应该能“消除”任何困惑：“一种根本不需要想象力的情景是：我们把军事武器的作战控制权交给人工智能模型，然后允许这些相同的人工智能模型自行决定使用军事武器杀死人类，过程中不需要任何人类的批准。从人工智能到人工智能杀死我们所有人，这根本不需要任何想象力。”
For that third group, Maddow said, the Post’s reporting should “cure” any confusion: “One scenario that takes really no imagination at all is a scenario in which we give AI models operational control over military weapons, and then we allow those same AI models to use military weapons to kill humans on their own say-so without any human having to OK it along the way. That doesn’t take any imagination at all to get from AI to AI killing us all.”
查看此帖：《华盛顿邮报》马多：人工智能武器报道应“消除”人们对该技术可能“杀死我们所有人”的任何困惑。
View this post on The post Maddow: AI weapons report should ‘cure’ any confusion that the technology could ‘kill us all’