文 / 王不留（微信公众号：考研英语笔记）
2021年11月1号的清晨，来杯“经济学人浓香咖啡”，提神解困。
Artificial selection: how evolution could design future robots
What if an algorithm could mimic hundreds of millions of years of evolution? A team from Stanford University tried to do just that. Their results, published in Nature Communications, a journal, could suggest ways to improve artificial intelligence.
The researchers created virtual environments filled with hundreds of creepy crawlies to test the approach. The “unimals” were programmed to learn tasks, such as moving a ball or manoeuvring through obstacles, while continuously evolving. As in the animal kingdom, the fittest member would reproduce. The offspring, which overwrote the parent, would be a physical mutation involving one genetic tweak: an extra limb for stability or a different joint for flexibility, perhaps. After 10 generations the most successful unimals could master tasks twice as fast as their ancestors, despite having the same baseline intelligence. Like in nature, a robot could adapt faster if its body and artificial mind evolve in tandem.
Artificial selection: how evolution could design future robots
人工选择：如何通过进化，设计未来机器人
evolution 进化 ; 演变 ; 发展 ; 渐进
future 将来 ; 未来 ; 未来的事 ; 将来发生的事 ; 前景 ; 前途 ; 前程 ; 将来的 ; 未来的 ; 将来发生的
robots 机器人 ; 交通信号灯 ; robot的复数
What if an algorithm could mimic hundreds of millions of years of evolution? A team from Stanford University tried to do just that. Their results, published in Nature Communications, a journal, could suggest ways to improve artificial intelligence.
如果算法能够模拟数亿年的进化会怎样？斯坦福大学的一支团队试图做到这点。他们的研究成果发表在《自然通讯》杂志上，可以为改进人工智能提供参考。
What if 如果…怎么办
algorithm 算法 ; 计算程序
mimic 模仿 ; 像 ; 会模仿的人
evolution 进化 ; 演变 ; 发展 ; 渐进
tried 尝试 ; 试 ; 试图 ; 努力 ; 审判 ; 试用 ; 想要 ; 设法 ; 试做 ; 试验 ; 审理 ; 审讯 ; 经过试用的;经过考验的;证明了的 ; try的过去分词和过去式
suggest 建议 ; 提议 ; 推荐 ; 举荐 ; 使想到 ; 使认为 ; 表明
artificial intelligence 人工智能
王不留(wbliu85)注：
Nature Communications，《自然通讯》是多学科类开放获取期刊中的翘楚，致力于生物、物理、化学和地球科学等各领域的高质量研究。
其发表的论文对各领域的专家来说都意味着相关学科的重大突破。
The researchers created virtual environments filled with hundreds of creepy crawlies to test the approach. The “unimals” were programmed to learn tasks, such as moving a ball or manoeuvring through obstacles, while continuously evolving. As in the animal kingdom, the fittest member would reproduce. The offspring, which overwrote the parent, would be a physical mutation involving one genetic tweak: an extra limb for stability or a different joint for flexibility, perhaps. After 10 generations the most successful unimals could master tasks twice as fast as their ancestors, despite having the same baseline intelligence. Like in nature, a robot could adapt faster if its body and artificial mind evolve in tandem.
研究人员创造了充满数百只毛骨悚然的爬虫的虚拟环境，来推演这种进化。”unimals”是一个被编程为学习任务（比如移动球或穿越障碍物）且不断进化的研究对象。就像在动物王国一样，最适应者将会繁衍后代。继承父母基因的后代可能会有一个涉及基因调整的物理突变：一个额外的肢体保持稳定性，或者一个不同的关节保持灵活性。10代之后，尽管它们拥有相同的基本智力，最成功的“unimals”掌握任务的速度快于它们祖先的两倍。就像自然界一样，如果机器人的身体和人工思维同步进化，它的适应能力会更快。
filled with 充满
hundreds of 成百上千 ; 好几百
Creepy 毛骨悚然的
crawlies 爬虫
such as 例如 ; 像 ; 象…这样 ; 诸如…之类
a ball 圆球
animal kingdom 动物王国
most successful 达到目的 ; 有成效的 ; 获得成功的 ; 有成就的 ; successful的最高级
fast as 和…一样快，迅速，立即
in tandem 同时的；一前一后的
王不留(wbliu85)注：
请大家欣赏这份研究成果。
，时长00:17
麻省理工学院的《MIT Technology Review(麻省理工科技评论)》杂志也报道了此事。
文章提供了更为详细的实验成果视频。有兴趣的朋友可以进一步阅读。