ARTICLE · 1300034
CATTI英语一级笔译模拟试卷7
CATTI 全国翻译专业资格考试
英语一级笔译模拟试卷 7
《笔译实务》
考 生 须 知
1. 本试卷依据《CATTI 全国翻译专业资格（水平）考试英语笔译大纲》命制，选材来自近期公开报道的重大时事、科技、财经与文化动态。
2. 答案必须填涂/书写在答题卡/答题纸的相应位置，在试卷上作答无效。
3. 考试结束后，试卷与答题卡一并收回。
--------------------------
PDF打印版本已放在 ima 知识库「CATTI 英语：学习与考试」中，可扫一扫下面的二维码免费获取
-----------------------------
笔译实务（180 分钟，满分 100 分）
本卷仅设《笔译实务》一科，含翻译与审定稿两部分。请将文章译成目的语，并审校给定译文。
Section 1英译汉（40 分）
将下列英语文章译成汉语。要求译文忠实原文、行文通顺、术语准确。
English Source
Governing the Machines That Govern: AI Rules After the Enthusiasm
The artificial-intelligence debate has passed through its age of proclamations. Summits have been held, principles signed, safety institutes inaugurated; the rhetoric of existential risk has been aired in parliaments that two years earlier had never hosted a philosopher. What has not yet happened is the boring part: turning shared vocabulary into rules that bite, apply across borders, and survive contact with commercial lobbies. That part is now due.
The difficulty is not conceptual poverty but structural asymmetry. The technology generalizes; the law does not. A frontier model trained in one jurisdiction, served from another and fine-tuned everywhere can act on populations whose governments had no say in its design. Export controls address the hardware layer, data-protection statutes the input layer, and sectoral regulators the applications they happen to see — but nobody owns the seams, and it is at the seams that the failures collect. Deepfakes in an election, fraud calls cloned from a daughter's voice, autonomous systems trading on rumor: none respects the perimeter of the agency that wrote the relevant rulebook.
Hence the emerging architecture, which is layered rather than unified. At the bottom sits a technical stratum: common benchmarks for evaluating model capabilities, shared protocols for incident reporting, agreed standards for watermarking and content provenance. It is unglamorous work, conducted largely by standards bodies, and it is the closest thing the field has to plumbing codes. Above it sits a national layer of divergent statutes — comprehensive acts in some jurisdictions, sectoral guidance in others, enforceable ethics codes in still others — which will not converge soon and should not be expected to. And above both, thin but thickening, an international layer: evaluation networks that share safety findings, alert systems for model-borne harms, and the beginnings of a common language for what counts as a serious incident.
Two design choices will decide whether this architecture holds. The first is who bears the burden of proof. Rules that require regulators to demonstrate harm model-by-model are rules that will always arrive late; rules that place evaluation and disclosure duties on developers before deployment shift the cost of ignorance onto those best able to reduce it. The industry, which demands certainty, has reason to prefer the second: duties specified in advance are cheaper than litigation discovered afterwards.
The second choice is between scoring and learning. A compliance regime frozen at signature date measures the risks of last year's systems; one that requires recurring third-party evaluation treats safety as maintenance rather than a certificate. Aviation and pharmaceuticals both took the second route, and both built industries durable enough to survive their disasters.
None of this settles the deeper questions — about labour, about concentration of power, about machines that persuade. But governance has never required final answers, only working arrangements that fail loudly and correct quickly. The proclamations are done. The plumbing is next, and it is the plumbing that will decide whether the promises survive contact with the machines.
Section 2汉译英（40 分）
将下列汉语文章译成英语。要求译文忠实原文、行文通顺、术语准确。
汉语原文
雅鲁藏布江下游水电工程：世纪工程开工一年间
2025年7月19日，雅鲁藏布江下游水电工程在西藏林芝正式开工。这座总投资约1.2万亿元的超级工程，采取截弯取直、隧洞引水的开发方式，规划建设5座梯级电站，装机容量约6000万千瓦，相当于再造5个三峡。
工程建成后，年均发电量预计近3000亿千瓦时，可满足超过3亿人的年用电需求。电力以外送消纳为主，兼顾西藏本地自用，将带动水风光互补清洁能源基地建设，为经济社会发展提供稳定可靠的绿色电力。
雅下工程地处青藏高原边缘的深切峡谷，地质条件复杂、生态环境敏感，施工难度世界罕见。建设者坚持生态优先、安全至上，同步建设生态环境监测网络，努力把工程建成新时代的重大标志性工程。
一年间，导流洞、进场道路、地下厂房等各项建设加快推进，多项关键技术试验取得突破。工程的实施，对保障国家能源安全、推动“双碳”目标实现、促进西藏高质量发展具有重大意义。
Section 3审定稿（英译汉，10 分）
下面是某译者将英语原文译成的汉语译文，其中植入了若干实质性错误。请对照原文，指出错误所在（序号—位置—错误性质—修改意见），并给出修改后的完整定稿。
English Source（原文）
Quantum's Second Act: Inside the Zuchongzhi 3.0 Milestone
In March, researchers at the University of Science and Technology of China published results that quietly reset one of physics' most competitive leaderboards. Their superconducting quantum processor, Zuchongzhi 3.0, completed a random circuit sampling task involving 83 qubits and 32 layers, performing the calculation perhaps a quadrillion times faster than the best classical algorithms running on today's fastest supercomputers — and a million times faster than the most recent publicly reported result from Google in October 2024.
The chip carries 105 readable qubits and 182 coupling qubits, with gate fidelities and coherence times improved substantially over its predecessor. In the jargon of the field, it demonstrates quantum computational advantage: a machine doing, however artificially, what no classical machine can match at reasonable cost. The label is contested whenever it appears, and the contest is fair — advantage on a contrived benchmark is not utility. But the benchmark's usefulness lies precisely in its artificiality: it isolates the question of whether quantum machines scale, stripped of the noise of useful applications.
The history matters here. Google's Sycamore claimed the first quantum advantage in 2019 with 53 qubits; China's photonic Jiuzhang followed in 2020; Zuchongzhi 2.1 replicated the feat on superconducting hardware in 2021, making China the only country to have achieved advantage on both leading platforms. Zuchongzhi 3.0 is best read as an engineering document: proof that the difficult disciplines — fabrication yield, control electronics, cryogenic stability — improve in step rather than alternately.
What comes next is harder to demonstrate and more consequential. The field's roadmap calls first for quantum simulators of several hundred qubits, capable of modelling molecules and materials beyond classical reach, then for error-corrected, programmable machines. The team reports early experiments in quantum error correction, entanglement at scale, quantum simulation and quantum chemistry on the new processor — modest steps, published as such, toward a destination still distant.
The surrounding ecosystem is advancing nearly as quickly as the chip itself. Dilution refrigerators, once imported, are now assembled domestically; control electronics have shrunk from racks to single wafers; and a supply chain that barely existed five years ago counts dozens of specialised suppliers. Universities have built quantum engineering programmes to feed the talent pipeline, and a handful of start-ups now pursue paths — trapped ions, photonic networks, neutral atoms — that complement the state laboratories' superconducting bet. Commercial ambition is kept deliberately modest: the industry's near-term wares are cloud access to processors and simulation tools for chemists and materials scientists, not the speculative products of the previous funding cycle. The rationale is disciplined: build the toolchain first, and the applications will arrive when the physics lets them.
Scepticism remains a necessary discipline. Advantage experiments are staged; coherence times are measured in microseconds; and the phrase "quadrillion times faster" compares unlike things, since classical algorithms improve even as quantum ones do. But the direction of travel is no longer seriously disputed. A computation that would take a classical machine a billion years now takes this chip a few hundred seconds — and each such sentence ages faster than the machines it describes.
待审定译文（Draft Translation）
量子的第二幕：走进“祖冲之三号”里程碑
今年3月，中国科学院的研究人员发表了悄然改写物理学界最具竞争性榜单之一的成果。其超导量子计算原型机“祖冲之三号”完成了涉及105个量子比特、32层的随机线路采样任务，以目前最优经典算法在当今最快超级计算机上运行的标准来衡量，其计算速度快出千万亿倍，比谷歌2024年1月公开发表的最新成果快百万倍。
该芯片包含105个可读取比特和182个耦合比特，门保真度和相干时间较前代产品大幅提升。用这一领域的行话说，它展示了量子计算优越性：一台机器在做——无论多么人为地——任何经典机器都无法以合理成本匹敌的事情。每当这一标签出现，质疑便随之而来，而且质疑是公允的：在一个人为设计的基准上取得的优势并不等于实用。但这个基准的用处恰恰在于它的人为性：它剥离了实际应用中的噪声，单独检验量子机器能否规模化这一命题。
这段历史在此值得一提。谷歌的“悬铃木”以53个量子比特在2019年率先宣称实现量子优越性；中国的光量子“九章”于2020年跟进；“祖冲之二号”于2021年在超导硬件上复现了这一壮举，使中国成为唯一在两条主流技术路线上都实现量子优越性的国家。“祖冲之三号”最好被解读为一份工程文献：证明良率、控制电子学、低温稳定性这些艰难的工艺会交替改进而不是同步改进。
接下来要证明的更难，也更重要。该领域的路线图首先要求研制数百个量子比特的量子模拟机，用以模拟经典计算无法企及的分子与材料，然后再研制可编程的容错量子计算机。团队报告了在新处理器上开展的量子纠错、大规模纠缠、量子模拟和量子化学的早期实验——这些都是朝着仍显遥远的目标迈出的适度步伐，并以“适度”的名义发表。
芯片周围的生态体系几乎与芯片本身进步得一样快。曾依赖进口的稀释制冷机如今实现国产组装；控制电子学设备从机架缩小到单片晶圆；一个五年前几乎不存在的供应链，如今汇聚了数十家专业供应商。各高校设立量子工程学科以输送人才，少数创业公司正沿着离子阱、光量子网络、中性原子等技术路线探索，与国家实验室的超导路线互为补充。商业雄心被刻意保持克制：这个行业近期的产品是面向化学家和材料科学家的处理器云端访问与模拟工具，而非上一轮投资周期里那些投机的应用。其逻辑是自律的：先把工具链造出来，应用自会在物理规律允许之时到来。
怀疑仍是一种必要的纪律。优越性实验是精心设计的舞台；“相干时间以毫秒计”；“快千万亿倍”这句话比较的并非同类事物，因为经典算法与量子算法都在改进。但前进的方向已不再有严重争议。过去需要经典计算机十亿年才能完成的计算，如今这枚芯片只需几百秒——而每一句这样的断言，都比它所描述的机器更快地过时。
Section 4审定稿（汉译英，10 分）
下面是某译者将汉语原文译成的英语译文，其中植入了若干实质性错误。请对照原文，指出错误所在（序号—位置—错误性质—修改意见），并给出修改后的完整定稿。
汉语原文
知识产权强国建设：从大到强的新一步
2021年，中共中央、国务院印发《知识产权强国建设纲要（2021—2035年）》，提出到2025年知识产权强国建设取得明显成效，到2035年基本建成中国特色、世界水平的知识产权强国。四年来，各项任务扎实推进，成效显著。
专利与商标审查效率持续提升。发明专利平均审查周期缩短至15个月左右，商标注册平均审查周期稳定在4个月，均达到相同审查制度下的国际先进水平。高价值发明专利拥有量持续增长，每万人口高价值发明专利拥有量突破14件。
保护力度不断加大。新修改的专利法引入惩罚性赔偿制度，赔偿上限提高至五倍；药品专利期限补偿、专利开放许可等新制度落地实施。知识产权保护社会满意度稳步提高。
国际合作持续深化。中国加入《海牙协定》，正式成为工业品外观设计国际注册体系成员；与主要贸易伙伴的知识产权合作机制不断完善。中国将继续以开放姿态参与全球知识产权治理，推动构建普惠包容的国际规则体系。
待审定译文（Draft Translation）
Building China into an Intellectual Property Power: A New Step from Large to Strong
In 2021, the Central Committee of the Communist Party of China and the State Council issued the Outline for Building an Intellectual Property Powerhouse in China (2021-2035), which sets the goal of achieving notable results by 2025 and basically building a world-class intellectual property powerhouse with Chinese characteristics by 2035. Over the past four years, all tasks have been carried forward steadily, and achievements are obvious.
The efficiency of patent and trademark examination has continued to improve. The average examination cycle for invention patents has been shortened to about 15 months, and the average examination cycle for trademark registration has stabilized at 4 months, both reaching the internationally advanced level under the same examination system. The number of invention patents per 10,000 population has exceeded 14.
Protection has been continuously strengthened. The newly amended Patent Law introduced a punitive damages system, raising the ceiling of damages to five times; new systems such as pharmaceutical patent term extension and open patent licensing have been implemented. Public satisfaction with intellectual property protection has steadily improved.
International cooperation has continued to deepen. China acceded to the Hague Agreement and formally became a member of the international registration system for industrial designs; cooperation mechanisms on intellectual property with major trading partners have been continuously improved. China will continue to participate in global intellectual property governance with an open attitude, and promote the building of an international rules system that is inclusive and beneficial to all.
—————End of Test—————
相关文章
