英语阅读：应对人工智能浪潮的策略

Thore Graepel may have been the first human to be conquered by a superintelligence. In 2015, on his first day as a researcher at Google DeepMind, he was challenged to play against the earliest iteration of AlphaGo—a computer program developed by DeepMind that would prove so effective at the ancient-Chinese game of weiqi (or Go, as it is commonly known in the West) that it changed how humans play it, and then overturned the field of AI itself.

vanquished/ˈvæŋkwɪʃt/ v. 彻底击败，征服；defeat completely in a battle or competition

superintelligence/ˌsuːpərɪnˈtelɪdʒəns/ n. 超级智能；artificial intelligence that far surpasses human brainpower

iteration/ˌɪtəˈreɪʃn/ n. 迭代版本，更新版；a new version of a program or product

upended/ʌpˈendɪd/ v. 彻底颠覆，推翻；completely change or reverse the structure of

中文翻译：索尔·格拉佩尔或许是首位被超级智能所征服的人类。2015年，当他入职谷歌深度思维担任研究员的第一天，就被挑战与最早的阿尔法Go版本对弈。这款由深度思维研发的计算机程序，在精通中国古老的围棋后，不仅重塑了人类的棋艺，更彻底颠覆了整个人工智能领域。

When Graepel faced it, AlphaGo was just a baby project, as he put it to me, and he was an accomplished amateur player. But it still took him down. Then, the following year, AlphaGo—now fully developed—plowed through a number of human champions, ultimately crushing Lee Sedol, widely considered the best player in the world, with a match score of 4–1. This month marked the tenth anniversary of that victory.

baby/ˈbeɪbi/ adj. 初创的，初期的；in the early stage of development

accomplished/əˈkʌmplɪʃt/ adj. 技艺高超的，熟练的；highly skilled or proficient

take down/teɪk daʊn/ phr. 击败，战胜；defeat an opponent

plow through/plaʊ θruː/ phr. 轻松接连击败；defeat a series of opponents easily

crushing/ˈkrʌʃɪŋ/ v. 碾压，完胜；defeat thoroughly

tenth anniversary/ˈtenθ ˌænɪˈvɜːsəri/ n. 十周年纪念；the tenth year of an important event

中文翻译：格拉佩尔曾向我透露，当时阿尔法Go还处于“婴儿期”，而他本人也是一名技艺高超的业余棋手，但他依然败下阵来。随后一年，完全成熟的阿尔法Go横扫多位人类顶尖冠军，最终以4:1的比分击败公认的世界第一李世石。本月正是这一历史性胜利的十周年。

For decades, developing a program that plays Go at an elite level was an infamous problem in computer science. Many considered it unsolvable—far harder than developing a similar program for chess, in which the supercomputer DeepBlue beat the world champion in 1997. In Go, two players take turns positioning stones on a 19-by-19 grid, and their movements are relatively unrestricted. In chess, which has a far smaller grid, a rook can move only horizontally and a bishop only diagonally, but Go pieces can be placed on any open space. The number of possible Go positions is so high that it cannot be easily expressed in words; it is higher than the number of atoms in the observable universe, and orders of magnitude higher than the number of possible chess games. Today, the technical frameworks and approaches that allowed an algorithm to excel at this board game have translated fairly directly into bots that can write advanced code, help tackle open problems in mathematics, and replicate scientific discoveries from scratch.

elite/ɪˈliːt/ adj. 顶尖的，精英级的；of the highest level or quality

infamous/ˈɪnfəməs/ adj. 声名狼藉的，极难解决的；well-known for being difficult or problematic

unsolvable/ʌnˈsɒlvəbl/ adj. 无法解决的；impossible to solve

unrestricted/ˌʌnrɪˈstrɪktɪd/ adj. 不受限制的；not limited or controlled

orders of magnitude/ˈɔːdəz ɒv ˈmæɡnɪtjuːd/ phr. 数量级（差距极大）；a very large difference in quantity or scale

algorithm/ˈælɡərɪðəm/ n. 算法；a set of rules for a computer to solve problems

excel at/ɪkˈsel æt/ phr. 擅长，精通；be extremely good at

tackle/ˈtækl/ v. 处理，解决（难题）；deal with a difficult problem

replicate/ˈreplɪkeɪt/ v. 复现，复制；repeat or create a copy of

from scratch/frəm skrætʃ/ phr. 从零开始；starting with nothing

中文翻译：数十年来，开发能下围棋的程序一直是计算机科学界的“圣杯”。许多人认为这几乎不可能，难度远超国际象棋（1997年深蓝已击败人类）。围棋棋盘19x19，落子自由，其变化数量远超可观测宇宙原子总数，比国际象棋高出几个数量级。如今，让算法精通围棋的技术，已直接演化为能编写高级代码、解决数学难题、从零复现科学发现的工具。

Generative AI is living in AlphaGo’s shadow. Beyond the actual models, “conceptual things emerged from the whole AlphaGo experience which essentially entered the AI vocabulary,” Pushmeet Kohli, the vice president of science and strategic initiatives at Google DeepMind, told me. In many ways, Go and chess provide ideal templates for understanding how the AI boom has unfolded—and a guide for what it may yet wreak.

conceptual/kənˈseptʃuəl/ adj. 概念上的；relating to ideas or concepts

emerge/ɪˈmɜːdʒ/ v. 出现，兴起；appear or become known

template/ˈtemplət/ n. 模板，范式；a model or pattern for understanding

unfold/ʌnˈfəʊld/ v. 展开，发展；develop or happen gradually

wreak/riːk/ v. 造成（影响/后果）；cause a large effect or damage

中文翻译：生成式AI始终笼罩在阿尔法Go的阴影之下。谷歌DeepMind科学与战略副总裁普什米特·科利表示：“从阿尔法Go项目中衍生出的概念，已融入AI的通用语言。”围棋与国际象棋不仅是理解AI热潮如何爆发的范本，也是预测其未来影响的指南。

DeepMind’s innovation was to essentially pair two algorithms: one AI model to propose moves and a second model to judge whether a move is good or not, allowing the system to devote computational resources to planning sequences of moves most likely to result in victory. AlphaGo then played itself thousands of times, improving from every mistake through a training process known as reinforcement learning. Today’s frontier AI labs face an analogous problem: Large language models such as ChatGPT could spit out lucid sentences and paragraphs, but when they faced challenging tasks in computer science, physics, and other areas that would require a human to really think, chatbots had been stuck stumbling in the dark. That began to change in late 2024 with the advent of so-called reasoning models, an approach that now underlies all of the top bots from OpenAI, Google DeepMind, and Anthropic. And the idea behind these reasoning models “is surprisingly similar to AlphaGo,” as Noam Brown, a researcher at OpenAI, recently put it.

pair/peə(r)/ v. 组合，配对；combine two things together

devote/dɪˈvəʊt/ v. 投入（时间/资源）；use time or resources for a purpose

reinforcement learning/ˌriːɪnˈfɔːsmənt ˈlɜːnɪŋ/ n. 强化学习；an AI training method using trial and error

frontier/ˈfrʌntɪə(r)/ adj. 前沿的；leading or advanced

analogous/əˈnæləɡəs/ adj. 相似的，可类比的；similar in a meaningful way

lucid/ˈluːsɪd/ adj. 清晰易懂的；clearly expressed and easy to understand

stumbling in the dark/ˈstʌmblɪŋ ɪn ðə dɑːk/ phr. 盲目摸索；act without clear direction

advent/ˈædvent/ n. 出现，问世；the arrival of a new technology

underlie/ˌʌndəˈlaɪ/ v. 构成…的基础；be the basic cause of

中文翻译：DeepMind的创新在于结合两种算法：一个提议走法，另一个评判优劣，使系统能集中算力于获胜路径。阿尔法Go通过数万次自我对弈，在强化学习中进化。如今前沿AI实验室面临类似困境：ChatGPT等模型虽能生成流畅文本，但在需深度思考的理工科任务上表现笨拙。2024年底，推理模型的出现扭转了局面，现已成为OpenAI、DeepMind等顶尖模型的核心。OpenAI研究员诺姆·布朗指出，其思路与阿尔法Go惊人相似。

The intuition behind chatbot reasoning is to have AI models work out a solution step-by-step, using a scratch pad of sorts, and then evaluate steps along the way to change course or start over as needed—very much like the two-step approach used by AlphaGo. The training method for these reasoning chatbots is the same as well: reinforcement learning. An algorithm can play lots of games of Go or attempt to solve lots of difficult math problems, then learn from its mistakes when it loses or errs. Today’s best AI models “can be traced back to some degree to the AlphaGo work,” Graepel said.

intuition/ˌɪntjuˈɪʃn/ n. 核心理念，直觉逻辑；the underlying guiding principle

scratch pad/skrætʃ pæd/ n. 草稿本；(AI)推理临时思考区；a temporary space for thinking and intermediate steps

start over/stɑːt ˈəʊvə/ phr. 重新开始，从头再来；begin again

err/ɜː(r)/ v. 犯错，出错；make a mistake or be incorrect

trace back to/treɪs bæk tuː/ phr. 追溯到；originate from

中文翻译：聊天机器人推理的核心在于分步解决问题，利用类似“草稿纸”的机制边评估边调整，这与阿尔法Go的两步法如出一辙。训练方法同样是强化学习：算法通过大量对弈和做题，从失败中汲取教训。格拉佩尔认为，当今最先进的AI模型在一定程度上都可追溯到阿尔法Go的研究成果。

Perhaps the most crucial insight shared between AlphaGo and the chatbot-reasoning breakthrough is a twist on the AI industry’s central dogma, the “scaling laws.” Traditionally, AI companies improved their large language models by training them on more data and with more computing power. In the case of AlphaGo and reasoning models, researchers realized that they could scale another dimension: having the program devote more time and computing power to a task, akin to how harder problems typically take humans more time to solve. For bots, this meant planning more and longer sequences of moves or using more words to “reason” through a tough coding task. That wasn’t guaranteed. “It could happen that you give them more time and they spend more time just getting confused,” Kohli said.

crucial/ˈkruːʃl/ adj. 至关重要的；extremely important

twist/twɪst/ n. 创新转变，新突破；a new or changed version of an idea

dogma/ˈdɒɡmə/ n. 信条，教条；a widely accepted principle

scaling laws/ˈskeɪlɪŋ lɔːz/ n. 规模法则；rules about increasing model size/power

scale/skeɪl/ v. 拓展，提升规模；expand or increase in size/scope

dimension/daɪˈmenʃn/ n. 维度，方面；an aspect or feature of a situation

akin to/əˈkɪn tuː/ phr. 类似于；similar to

中文翻译：阿尔法Go与推理模型最关键的启示，在于革新了AI行业的“规模法则”信条。传统上，企业靠堆砌数据和算力提升性能。而两者让研究者发现新维度：让程序投入更多时间与算力解决任务，正如人类解难题需耗时。对机器人而言，意味着规划更长时间序列或用更多文字推理。但这未必奏效，科利警告：“给它更多时间，它可能只会更困惑。”

After the success of AlphaGo, DeepMind made a successor program called AlphaZero. Whereas AlphaGo was initially shown a number of human Go matches as a baseline, AlphaZero became dominant at a number of games—Go, chess, and so on—purely by playing itself, with zero prior knowledge, and learning from each game. That an AI model essentially taught itself, very rapidly, to surpass the abilities of any human ever at multiple games might suggest that very rapid advances for today’s chatbots are on the horizon. By this logic, models could essentially figure out ways to improve themselves. But the success of AlphaGo and AlphaZero more likely signals obstacles ahead. The most important ingredient in AlphaGo was the simplicity with which one could measure success—win or lose—and thus give the machine feedback to improve.

successor/səkˈsesə(r)/ n. 继任者，后续版本；a thing that follows another

baseline/ˈbeɪslaɪn/ n. 基准，基础；a starting point for comparison

dominant/ˈdɒmɪnənt/ adj. 占绝对优势的；most powerful or successful

prior knowledge/ˈpraɪə(r) ˈnɒlɪdʒ/ n. 先验知识；existing knowledge before training

surpass/səˈpɑːs/ v. 超越，超过；do better than

on the horizon/ɒn ðə həˈraɪzn/ phr. 即将来临；about to happen soon

obstacle/ˈɒbstəkl/ n. 障碍，困难；something that blocks progress

ingredient/ɪnˈɡriːdiənt/ n. 要素，因素；an essential part of something

simplicity/sɪmˈplɪsəti/ n. 简单性，简洁性；the quality of being easy to understand

feedback/ˈfiːdbæk/ n. 反馈；information returned to improve performance

中文翻译：阿尔法Go成功后，DeepMind推出了阿尔法零。不同于以人类棋谱为基准的阿尔法Go，阿尔法零完全从零开始，仅通过自我对弈就在围棋、国际象棋等领域占据统治地位。AI能快速自学超越人类，似乎预示ChatGPT等模型即将迎来爆发式突破。按此逻辑，模型甚至能自我优化。但阿尔法Go与阿尔法零的成功更可能预示前路有险阻。其关键在于成功标准极简——输赢，从而为机器提供改进反馈。

With board games, “we were always operating in a specific environment where the rules of the game were known,” Kohli said. “The systems of today are expected to operate in a much more general environment.” Reasoning models have found success mostly in areas that still have a relatively clear rubric for evaluation: whether an AI-written program works as intended, for instance, or whether an AI-written proof holds up. Instilling any notion of a more general intelligence in a machine will be a far more challenging problem than conquering even Go.

rubric/ˈruːbrɪk/ n. 评估标准；a set of rules for judging quality

evaluation/ɪˌvæljuˈeɪʃn/ n. 评估，评价；the act of judging value or quality

hold up/həʊld ʌp/ phr. 成立，站得住脚；prove to be true or valid

instill/ɪnˈstɪl/ v. 逐步灌输；gradually introduce an idea

conquer/ˈkɒŋkə(r)/ v. 攻克，征服；defeat or master completely

中文翻译：科利指出：“棋类游戏处于规则明确的封闭环境，而今日系统需在通用环境中运行。”推理模型目前多在评估标准清晰的领域成功，如程序能否运行、证明是否成立。要让机器具备通用智能，其难度远超攻克围棋。

DeepMind has been able to design evaluations for more abstract ideas, for instance by orchestrating several AI agents to act as a team of virtual “scientists” that will rank hypotheses about problems in biology. But even that system operates within a relatively constrained domain of biological reasoning and literature. It’s unlikely that any lab will come up with a single way to evaluate “general intelligence” that can be used to train a bot AlphaGo style, let alone one as straightforward as winning or losing a board game.

orchestrate/ˈɔːkɪstreɪt/ v. 协调，统筹；organize and manage carefully

hypothesis/haɪˈpɒθəsɪs/ n. 假设；a proposed explanation for a problem

constrained/kənˈstreɪnd/ adj. 受限的；limited or restricted

domain/dəˈmeɪn/ n. 领域，范围；a particular area of knowledge

come up with/kʌm ʌp wɪð/ phr. 提出，想出；invent or suggest a method

let alone/let əˈləʊn/ phr. 更不用说；not to mention

straightforward/ˌstreɪtˈfɔːwəd/ adj. 简单直接的，易懂的；easy to understand

中文翻译：DeepMind已能评估抽象概念，例如让多个AI智能体扮演“科学家”对生物学假设排序。但这仍局限于特定领域。任何实验室都难以提出像阿尔法Go那样能训练通用智能的评估法，更别提像棋类胜负那样简单的标准。

Still, the progress the AlphaGo approach has yielded for AI models in a number of scientific domains is impressive—so much so that, a decade after AI conquered humanity’s hardest board game, the nation is now in a frenzy over whether AI is about to first overhaul the economy and then unsettle the purpose of being human at all.

yield/jiːld/ v. 产生，带来；produce a result or progress

domain/dəˈmeɪn/ n. 领域；an area of activity or study

in a frenzy/ɪn ə ˈfrenzi/ phr. 处于狂热中；in a state of intense excitement

overhaul/ˈəʊvəhɔːl/ v. 彻底改革；completely change or improve

unsettle/ʌnˈsetl/ v. 动摇，使不安；disturb or upset

中文翻译：尽管如此，阿尔法Go的技术路线在多个科学领域带来了巨大进步。AI攻克人类最难棋类游戏十周年之际，全社会陷入狂热：AI是否将彻底改革经济，甚至动摇人类存在的意义？

Once again, chess and Go might offer guides. As a result of improving via self-play, AlphaGo and AlphaZero developed not only superhuman ability but also inhuman style, using tactics and strategies no human had previously considered. These AI strategies did not destroy the human pursuits of chess and Go; they reignited new waves of human creativity and strategy. The most optimistic analogy for today’s more broadly useful AI systems would be that they also, rather than providing a wholesale replacement for humans, will function as a sort of complementary intelligence. Biologists, mathematicians, and computer scientists are already finding ways in which today’s AI models are not simply speeding up their work but qualitatively changing the kinds of questions humans can ask and the discoveries we can make.

reignite/ˌriːɪɡˈnaɪt/ v. 重新点燃，重启；start again with new energy

optimistic/ˌɒptɪˈmɪstɪk/ adj. 乐观的；positive and hopeful

analogy/əˈnælədʒi/ n. 类比，比喻；a comparison between similar things

wholesale/ˈhəʊlseɪl/ adj. 大规模的，全盘的；complete or total

complementary intelligence/ˌkɒmplɪˈmentri ɪnˈtelɪdʒəns/ n. 互补智能；intelligence that supports humans

qualitatively/ˈkwɒlɪtətɪvli/ adv. 本质地，质变地；in terms of quality rather than quantity

中文翻译：国际象棋与围棋再次提供了启示。通过自我对弈，阿尔法Go与阿尔法零不仅拥有超人类能力，还演化出人类未曾设想的战术。这些AI并未摧毁人类棋类运动，反而激发了新的创造力与策略。对通用AI最乐观的类比是：它们不会全盘取代人类，而是作为互补智能存在。生物学家、数学家等已发现，AI不仅加速工作，更从本质上改变了人类能提出的问题与发现的广度。

Of course, the business proposition of generative AI is quite the opposite: that products such as ChatGPT and Claude Code can automate huge swathes of white-collar work, help students cheat their way through school, and allow humans to live mostly without thinking. Perhaps C-suite executives, like AI researchers, can learn a lesson from Go and chess. Like any sport, chess and Go are worthwhile because of human struggles and storylines, champions made and toppled, the very fact that people are doomed to be imperfect but always striving to become just a bit better. And rather than automating human chess masters or destroying the sport and pastime, chess-playing AI models have helped the business of chess to boom.

proposition/ˌprɒpəˈzɪʃn/ n. 理念，主张；a business or value idea

automate/ˈɔːtəmeɪt/ v. 使自动化；make a process operate by machines

swaths/swɒðz/ n. 大片，大部分；large parts or areas

cheat their way through/tʃiːt weɪ θruː/ phr. 作弊混过；pass by dishonest methods

worthwhile/ˌwɜːθˈwaɪl/ adj. 有价值的，值得的；valuable and deserving effort

topple/ˈtɒpl/ v. 推翻，使下台；remove from a top position

are doomed to/ɑː duːmd tu/ phr. 注定；be certain to do or be something

pastime/ˈpɑːstaɪm/ n. 消遣，娱乐；an activity done for pleasure

boom/buːm/ v. 繁荣，迅速发展；grow quickly and successfully

中文翻译：当然，生成式AI的商业理念截然不同：ChatGPT等产品可自动化大量白领工作、助长学生作弊、让人无需思考。或许高管和研究者能从棋类中学到一课：棋类价值在于人类的拼搏、故事与冠军更迭；人类虽不完美却追求进步。下棋AI并未取代人类棋手，反而推动了棋类商业的繁荣。

Likewise, employees, managers, students, professors—really all of us—are always learning and learning by failing, or at least we should be. That is useful and worth preserving in plain economic terms. Nobody becomes world-class at anything without at some point being rather terrible at it, and allowing novices who might be less capable than a bot to build up skills is the only way you get experts with human judgment and abilities that surpass any AI. But more important than that economic rationale is an existential one: To grow or help another do so is a beautiful thing. Some might call it being human.

preserve/prɪˈzɜːv/ v. 保留，维护；keep and maintain

novice/ˈnɒvɪs/ n. 新手，初学者；a person new to a skill or activity

surpass/səˈpɑːs/ v. 超越；do better than

rationale/ˌræʃəˈnɑːl/ n. 根本理由，逻辑；the fundamental reason for something

existential/ˌeɡzɪˈstenʃl/ adj. 存在的，关乎生命本质的；relating to human existence

中文翻译：同样，员工、管理者、学生——我们所有人，都在不断学习、在失败中成长。从经济角度看，这是有价值的。没人能不经笨拙期成为专家；允许新手（甚至不如AI的）积累技能，是培养有人类判断力专家的唯一途径。但比经济逻辑更重要的是存在价值：自我成长与助人成长是美好的，这就是人性。