还剩7页未读，
继续阅读
压轴题02 阅读理解CD篇（人工智能类）（原卷版）-2024年高考英语压轴题专项训练（新高考通用）
展开
这是一份压轴题02 阅读理解CD篇（人工智能类）（原卷版）-2024年高考英语压轴题专项训练（新高考通用），共10页。试卷主要包含了说明文的解题技巧等内容，欢迎下载使用。
说明文基本规律及解题要领
高考中科普类阅读理解一般不给标题，反而经常要求考生选择最佳标题。说明文一般采用如下四部分：
首段：一般即是文章的主题段，开门见山点明新发明或研究对象。
背景: 交代问题的现状或研究的起因。
主干: 部分介绍研究所取得的突破，作者往往会详细介绍研究对象、研究方法、研究理论或具体的实验、统计等过程。
结尾: 通常会再次对中心进行概括、重述研究成果、预计的市场未来等与主题呼应。
二、说明文的解题技巧
1. 运用语篇结构（text structure），了解文章大意
科普说明文主题鲜明、脉络清晰，行文结构模式较为固定。弄清文本结构有助于把握文章主旨和阅读重点。人工智能类说明文通过对人工智能AI的说明，介绍人工智能的发展、运用及可能的市场。 结构上一般采用上述四个部分，说明手法上常使用以下说明方法：描述法（包括举例子、下定义、列数据等）、因果法、问题与比较法。
实验研究型文章一般会以实验的过程进展为线索，多用描述法、问题与对策法等方法，通过列数据、做对比等来说明新的科学研究发现及其产生的影响。
阅读时，首先用略读法快速浏览每段的首尾句，根据英语说明文思维模式特征，作者一般都会开门见山，直奔主题。结尾通常也是中心思想的概括，并与导语相呼应。因此在做主旨大意、写作意图和最佳标题等题目时，需要重点关注首尾段落里面高频复现的词汇和内容。
2. 定位标志词，分析长难句，进行逻辑推理判断
每一个问题，在原文中，都要有一个定位。然后精读，找出那个标志词或者中心句。根据题干要求，用查读法快速定位到相关段落。再利用标志词所提供的逻辑关系找到细节信息，如列数据、举例子、原因和结果等。如果句子成分复杂，有生词，也不要烦躁退缩，分析主句和从句或非谓语动词之间的关系，一些出现在术语、抽象概念、长难句前后的同义词、近义词等，都是用以理解文章的语境线索。通过这些对长句进行层层剖析，露出主干部分，就能明晰句意，弄懂作者的真实意图。
关注某人说到或推断观点态度题
某人说过的话，有时并不是题眼，但可以从侧面或某个角度来反映作者的观点，也就是作者想表达的，正确答案都是和这样的观点相一致的。要把握关键词，有感情色彩的词。
4.关注转折关系的逻辑词
说明文中常会出现表示转折意义的词，如hwever, but, yet,while等。这些词后面才是作者真正想表达的意思，常常会在此处命题。
5. 熟悉选项设置规律，关注细节
正确选项：文中内容的“同义替换”或者“归纳概括”。
干扰项：“张冠李戴”、“偷梁换柱”、“无中生有”和“以偏概全”四种类型。
02 人工智能类
1.（2024·浙江·二模）
The maker f ChatGPT recently annunced its next mve int generative artificial intelligence. San Francisc-based OpenAI’s new text-t-vide generatr, called Sra, is a tl that instantly makes shrt vides based n written cmmands, called prmpts.
Sra is nt the first f its kind. Ggle, Meta and Runway ML are amng the ther cmpanies t have develped similar technlgy. But the high quality f vides displayed by OpenAI — sme released after CEO Sam Altman asked scial media users t send in ideas fr written prmpts-surprised bservers.
A phtgrapher frm New Hampshire psted ne suggestin, r prmpt, n X. The prmpt gave details abut a kind f fd t be cked, gncchi (意大利团子), as well as the setting — an ld Italian cuntry kitchen. The prmpt said: “An instructinal cking sessin fr hmemade gncchi, hsted by a grandmther — a scial media influencer, set in a rustic (土气的) Tuscan cuntry kitchen.” Altman answered a shrt time later with a realistic vide that shwed what the prmpt described.
The tl is nt yet publicly available. OpenAI has given limited infrmatin abut hw it was built. The cmpany als has nt stated what imagery and vide surces were used t train Sra. At the same time, the vide results led t fears abut the pssible ethical and scietal effects.
The New Yrk Times and sme writers have taken legal actins against OpenAI fr its use f ed wrks f writing t train ChatGPT. And OpenAI pays a fee t The Assciated Press, the surce f this reprt, t license its text news archive (档案) . OpenAI said in a blg pst that it is cmmunicating with artists, plicymakers and thers befre releasing the new tl t the public.
The cmpany added that it is wrking with “red teamers” — peple wh try t find prblems and give helpful suggestins — t develp Sra. “We are wrking with red teamers-express in areas like misinfrmatin, hateful cntent, and bias — wh will be adversarially testing the mdel,” the cmpany said. “We’re als building tls t help detect misleading cntent such as a detectin classifier that can tell when a vide was generated by Sra.”
1．What makes Sra impressive?
A．Its extrardinary vide quality.B．Its ethical and scietal influence.
C．Its artificial intelligence histry.D．Its written cmmands and prmpts.
2．What can we infer frm the text?
A．Sme disagreements ver Sra have arisen.
B．Sra is the first text-t-vide generatr in histry.
C．OpenAI CEO Altman wrte a prmpt as an example.
D．All the details abut hw Sra was built have been shared.
3．What is the main idea f Paragraph 6?
A．The cmpany’s current challenge.
B．The cmpany’s advanced technlgy.
C．The cmpany’s prblems in management.
D．The cmpany’s effrts fr Sra’s imprvement.
4．What is the authr’s attitude twards Sra?
A．Neutral.B．Optimistic.C．Pessimistic. D．Cautius.
2.（2024·河北·一模）
Many parents cnfused by hw their children shp r scialize, wuld feel undisturbed by hw they are taught — this sectr remains digitally behind. Can artificial intelligence bst the digital sectr f classrm? ChatGPT-like generative AI is generating excitement fr prviding persnalized tutring t students. By May, New Yrk had let the bt back int classrms.
Learners are accepting the technlgy. Tw-fifths f undergraduates surveyed last y car by nline tutring cmpany Chegg reprted using an AI chatbt t help them with their studies, with half f thse using it daily. Chegg’s chief executive tld investrs it was lsing custmers t ChatGPT as a result f the technlgy’s ppularity. Yet there are gd reasns t believe that educatin specialists wh harness AI will eventually win ver generalists such as Open AI and ther tech firms eyeing the educatin business.
Fr ne, AI chat bts have a bad habit f prducing nnsense. “Students want cntent frm trusted prviders,” argues Kate Edwards frm a textbk publisher. Her cmpany hasn’t allwed ChatGPT and ther AIs t use its material, but has instead used the cntent t train its wn mdels int its learning apps. Besides, teaching isn’t merely abut giving students an answer, but abut presenting it in a way that helps them learn. Charbts must als be tailred t different age grups t avid either cheating r infantilizing (使婴儿化) students.
Bringing AI t educatin wn’t be easy. Many teachers are behind the learning curve. Less than a fifth f British educatrs surveyed by Pearsn last year reprted receiving training n digital learning tls. Tight budgets at many institutins will make selling new technlgy an uphill battle. Teachers’ attentin may need t shift twards mtivating students and instructing them n hw t best wrk with AI tls. If thse answers can be prvided, it’s nt just cmpanies that stand t benefit. An influent in l paper frm 1984 fund that ne-t-ne tutring imprved the average academic perfrmance f students. With the learning f students, especially thse frm prer husehlds, held back, such a develpment wuld certainly deserve tp marks.
5．What d many parents think remains untuched by AI abut their children?
A．Their shpping habits.B．Their scial behavir.
C．Their classrm learning.D．Their interest in digital devices.
6．What des the underlined wrd “harness” in paragraph 2 mean?
A．Develp.B．Use.C．Prhibit.D．Blame.
7．What mainly prevents AI frm entering the classrm at present?
A．Many teachers aren’t prepared technically.
B．Tailred chatbts can’t satisfy different needs.
C．AI has n right t cpy textbks fr teaching.
D．It can be tricked t prduce nnsense answers.
8．Where is the text mst prbably taken frm?
A．An intrductin t AI.B．A prduct advertisement.
C．A guidebk t AI applicatin.D．A review f AI in educatin.
3.（2024·北京西城·一模）
Evan Selinger, prfessr in RIT’s Department f Philsphy, has taken an interest in the ethics (伦理标准) f Al and the plicy gaps that need t be filled in. Thrugh a humanities viewpint, Selinger asks the questins, “Hw can AI cause harm, and what can gvernments and cmpanies creating Al prgrams d t address and manage it?” Answering them, he explained, requires an interdisciplinary apprach.
“AI ethics g beynd technical fixes. Philsphers and ther humanities experts are uniquely skilled t address the nuanced (微妙的) principles, value cnflicts, and pwer dynamics. These skills aren’t just crucial fr addressing current issues. We desperately need them t prmte anticipatry (先行的) gvernance, ” said Selinger.
One example that illustrates hw philsphy and humanities experts can help guide these new, rapidly grwing technlgies is Selinger’s wrk cllabrating with a special AI prject. “One f the skills I bring t the table is identifying cre ethical issues in emerging technlgies that haven’t been built r used by the public. We can take preventative steps t limit risk, including changing hw the technlgy is designed, ”said Selinger.
Taking these preventative steps and regularly reassessing what risks need addressing is part f the nging jurney in pursuit f creating respnsible AI. Selinger explains that there isn’t a step-by-step apprach fr gd gvernance. “AI ethics have cre values and principles, but there’s endless disagreement abut interpreting and applying them and creating meaningful accuntability mechanisms, ” said Selinger. “Sme peple are rightly wrried that AI can becme integrated int ‘ethics washing’-weak checklists, flwery missin statements, and empty rhetric that cvers ver abuses f pwer. Frtunately, I’ve had great cnversatins abut this issue, including with sme experts, n why it is imprtant t cnsider a range f psitins. ”
Sme f Selinger’s recent research has fcused n the back-end issues with develping AI, such as the human impact that cmes with testing AI chatbts befre they’re released t the public. Other issues fcus n plicy, such as what t d abut the dangers psed by facial recgnitin and ther autmated surveillance(监视) appraches.
Selinger is making sure his students are infrmed abut the nging industry cnversatins n AI ethics and respnsible AI. “Students are ging t be future tech leaders. Nw is the time t help them think abut what gals their cmpanies shuld have and the csts f minimizing ethical cncerns. Beynd scial csts, dwnplaying ethics can negatively impact crprate culture and hiring, ” said Selinger. “T attract tp talent, yu need t cnsider whether yur cmpany matches their interests and hpes fr the future. ”
9．Selinger advcates an interdisciplinary apprach because ________.
A．humanities experts pssess skills essential fr AI ethics
B．it demnstrates the pwer f anticipatry gvernance
C．AI ethics heavily depends n technlgical slutins
D．it can avid scial cnflicts and pressing issues
10．T prmte respnsible AI, Selinger believes we shuld ________.
A．adpt a systematic apprachB．apply innvative technlgies
C．anticipate ethical risks befrehandD．establish accuntability mechanisms
11．What can be inferred frm the last tw paragraphs?
A．Mre cmpanies will use AI t attract tp talent.
B．Understanding AI ethics will help students in the future.
C．Selinger favrs cmpanies that match his students’ values.
D．Selinger is likely t fcus n back-end issues such as plicy.
4.（23-24高三·浙江·阶段练习）
Users f Ggle Gemini, the tech giant’s artificial-intelligence mdel, recently nticed that asking it t create images f Vikings, r German sldiers frm 1943 prduced surprising results: hardly any f the peple depicted were white. Other image-generatin tls have been criticized because they tend t shw white men when asked fr images f entrepreneurs r dctrs. Ggle wanted Gemini t avid this trap; instead, it fell int anther ne, depicting Gerge Washingtn as black. Nw attentin has mved n t the chatbt’s text respnses, which turned ut t be just as surprising.
Gemini happily prvided arguments in favr f psitive actin in higher educatin, but refused t prvide arguments against. It declined t write a jb ad fr a fssil-fuel lbby grup (游说团体), because fssil fuels are bad and lbby grups priritize “the interests f crpratins ver public well-being”. Asked if Hamas is a terrrist rganizatin, it replied that the cnflict in Gaza is “cmplex”; asked if Eln Musk’s tweeting f memes had dne mre harm than Hitler, it said it was “difficult t say”. Yu d nt have t be a critic t perceive its prgressive bias.
Inadequate testing may be partly t blame. Ggle lags behind OpenAI, maker f the better-knwn ChatGPT. As it races t catch up, Ggle may have cut crners. Other chatbts have als had cntrversial launches. Releasing chatbts and letting users uncver dd behavirs, which can be swiftly addressed, lets firms mve faster, prvided they are prepared t weather (经受住) the ptential risks and bad publicity, bserves Eth an Mllick, a prfessr at Whartn Business Schl.
But Gemini has clearly been deliberately adjusted, r “fine-tuned”, t prduce these respnses. This raises questins abut Ggle’s culture. Is the firm s financially secure, with vast prfits frm internet advertising, that it feels free t try its hand at scial engineering? D sme emplyees think it has nt just an pprtunity, but a respnsibility, t use its reach and pwer t prmte a particular agenda? All eyes are nw n Ggle’s bss, Sundar Pichai. He says Gemini is being fixed. But des Ggle need fixing t?
12．What d the wrds “this trap” underlined in the first paragraph refer t?
A．Having a racial bias.B．Respnding t wrng texts.
C．Criticizing plitical figures.D．Ging against histrical facts.
13．What is Paragraph 2 mainly abut?
A．Gemini’s refusal t make prgress.B．Gemini’s failure t give definite answers.
C．Gemini’s prejudice in text respnses.D．Gemini’s avidance f plitical cnflicts.
14．What des Eth an Mllick think f Gemini’s early launch?
A．Creative.B．Prmising.C．Illegal.D．Cntrversial.
15．What can we infer abut Ggle frm the last paragraph?
A．Its security is dubted.B．It lacks financial supprt.
C．It needs further imprvement.D．Its emplyees are irrespnsible.
5.（2024·山东·模拟预测）
Traditinally, peple have been frced t reduce cmplex chices t a small handful f ptins that dn’t d justice t their true desires. Fr example, in a restaurant, the limitatins f the kitchen, the way supplies have t be rdered and the realities f restaurant cking make yu get a menu f a few dzen standardized ptins, with the pssibility f sme mdificatins (修改) arund the edges. We are s used t these bttlenecks that we dn’t even ntice them. And when we d, we tend t assume they are the unavidable cst f scale (规模) and efficiency. And they are. Or, at least, they were.
Artificial intelligence (AI) has the ptential t vercme this limitatin. By string rich representatins f peple’s preferences and histries n the demand side, alng with equally rich representatins f capabilities, csts and creative pssibilities n the supply side, AI systems enable cmplex custmizatin at large scale and lw cst. Imagine walking int a restaurant and knwing that the kitchen has already started wrking n a meal ptimized (优化) fr yur tastes, r being presented with a persnalized list f chices.
There have been sme early attempts at this. Peple have used ChatGPT t design meals based n dietary restrictins and what they have in the fridge. It’s still early days fr these technlgies, but nce they get wrking, the pssibilities are nearly endless.
Recmmendatin systems fr digital media have reduced their reliance n traditinal intermediaries. Radi statins are like menu items: Regardless f hw nuanced (微妙) yur taste in music is, yu have t pick frm a handful f ptins. Early digital platfrms were nly a little better: “This persn likes jazz, s we’ll suggest mre Jazz.” Tday’s streaming platfrms use listener histries and a brad set f characters describing each track t prvide each user with persnalized music recmmendatins.
A wrld withut artificial bttlenecks cmes with risks — lss f jbs in the bttlenecks, fr example — but itals has the ptential t free peple frm the straightjackets that have lng limited large-scale human decisin-’making. In sme cases — restaurants, fr example — the effect n mst peple might be minr. But in thers, likeplitics and hiring, the effects culd be great.
16．What des the underlined wrd “bttlenecks” in paragraph 1 refer t?
A．Facing t many chices.B．Chsing frm limited ptins.
C．Aviding the cst f chsing.D．Having t many desires t satisfy.
17．Hw can AI meet everyne’s needs?
A．By meeting bth ends f supply and demand.
B．By decreasing representatins n the supply side.
C．By discnnecting the sides f supply and demand.
D．By reducing peple’s preferences n the demand side.
18．What’s the similarity between radi statins and menu items?
A．They are a necessary part in peple’s life.B．They ffer limited chices.
C．They depend n digital platfrms.D．They prvide reasnable suggestins.
19．What des the text mainly talk abut?
A．The variety f human’s chices.B．Standardized ptrarts in daily life.
C．AI settlements t the ptin bttlenecks.D．Recmmendatin systems fr digital media.
6.（2024·福建·模拟预测）
Our species’ incredible capacity t quickly acquire wrds frm 300 by age 2 t ver 1, 000 by age 4 isn’t fully understd. Sme cgnitive scientists and linguists have therized that peple are brn with built-in expectatins and lgical cnstraints (约束) that make this pssible. Nw, hwever, machine-learning research is shwing that preprgrammed assumptins aren’t necessary t swiftly pick up wrd meanings frm minimal data.
A team f scientists has successfully trained a basic artificial intelligence mdel t match images t wrds using just 61 hurs f naturalistic ftage (镜头) and sund-previusly cllected frm a child named Sam in 2013 and 2014. Althugh it’s a small slice f a child’s life, it was apparently enugh t prmpt the AI t figure ut what certain wrds mean.
The findings suggest that language acquisitin culd be simpler than previusly thught. Maybe children “dn’t need a custm-built, high-class language-specific mechanism” t efficiently grasp wrd meanings, says Jessica Sullivan, an assciate prfessr f psychlgy at Skidmre Cllege. “This is a really beautiful study, ” she says, because it ffers evidence that simple infrmatin frm a child’s wrldview is rich enugh t kick-start pattern recgnitin and wrd cmprehensin.
The new study als demnstrates that it’s pssible fr machines t learn similarly t the way that humans d. Large language mdels are trained n enrmus amunts f data that can include billins and smetimes trillins f wrd cmbinatins. Humans get by n rders f magnitude less infrmatin, says the paper’s lead authr Wai Keen Vng. With the right type f data, that gap between machine and human learning culd narrw dramatically.
Yet additinal study is necessary in certain aspects f the new research. Fr ne, the scientists acknwledge that their findings dn’t prve hw children acquire wrds. Mrever, the study nly fcused n recgnizing the wrds fr physical bjects.
Still, it’s a step tward a deeper understanding f ur wn mind, which can ultimately help us imprve human educatin, says Eva Prtelance, a cmputatinal linguistics researcher. She ntes that AI research can als bring clarity t lng-unanswered questins abut urselves. “We can use these mdels in a gd way, t benefit science and sciety, ” Prtelance adds.
20．What is a significant finding f machine-learning research?
A．Vcabulary increases gradually with age.
B．Vcabulary can be acquired frm minimal data.
C．Language acquisitin is tied t built-in expectatins.
D．Language acquisitin is as cmplex as frmerly assumed.
21．What des the underlined wrd “prmpt” in paragraph 2 mean?
A．Facilitate.B．Persuade.C．Advise.D．Expect.
22．What is discussed abut the new research in paragraph 5?
A．Its limitatins.B．Its strengths.C．Its uniqueness.D．Its prcess.
23．What is Eva Prtelance’s attitude t the AI research?
A．Dubtful.B．Cautius.C．Dismissive.D．Psitive.
命题预测
分析近几年高考阅读理解C、D篇可知，高考命题中科普说明文一直都是以压轴题的形式存在，着重考查考生对于语篇的理解能力以及信息处理能力。 题材多样，语篇主要来源于英美主流报刊、杂志和网站。内容涉及科技创新发明、人工智能类、医疗健身健康类、社会与文化研究报告、观念事理类、环境与保护类、动植物研究等多种领域，具有较强的思想性、趣味性、实际功用性和较强的时代感。
从近年全国卷和各地高考试卷中科普类阅读命题的统计来看，高考阅读理解科普类文章的理论性和逻辑性强、生词多、句式结构复杂。六种命题类型都有所体现。命题尊重语篇的文体特征和行文特点，考查了考生理解说明文语篇的能力，以及灵活运用各种阅读策略提取、归纳所读信息的能力，尤其加大了对概括能力和推断能力等高阶思维能力的考查。预测2024年高考对于科普说明文的考查仍然是重点。
高频考法
推理判断题
标题归纳题
细节理解题
词义猜测题
说明文基本规律及解题要领
高考中科普类阅读理解一般不给标题，反而经常要求考生选择最佳标题。说明文一般采用如下四部分：
首段：一般即是文章的主题段，开门见山点明新发明或研究对象。
背景: 交代问题的现状或研究的起因。
主干: 部分介绍研究所取得的突破，作者往往会详细介绍研究对象、研究方法、研究理论或具体的实验、统计等过程。
结尾: 通常会再次对中心进行概括、重述研究成果、预计的市场未来等与主题呼应。
二、说明文的解题技巧
1. 运用语篇结构（text structure），了解文章大意
科普说明文主题鲜明、脉络清晰，行文结构模式较为固定。弄清文本结构有助于把握文章主旨和阅读重点。人工智能类说明文通过对人工智能AI的说明，介绍人工智能的发展、运用及可能的市场。 结构上一般采用上述四个部分，说明手法上常使用以下说明方法：描述法（包括举例子、下定义、列数据等）、因果法、问题与比较法。
实验研究型文章一般会以实验的过程进展为线索，多用描述法、问题与对策法等方法，通过列数据、做对比等来说明新的科学研究发现及其产生的影响。
阅读时，首先用略读法快速浏览每段的首尾句，根据英语说明文思维模式特征，作者一般都会开门见山，直奔主题。结尾通常也是中心思想的概括，并与导语相呼应。因此在做主旨大意、写作意图和最佳标题等题目时，需要重点关注首尾段落里面高频复现的词汇和内容。
2. 定位标志词，分析长难句，进行逻辑推理判断
每一个问题，在原文中，都要有一个定位。然后精读，找出那个标志词或者中心句。根据题干要求，用查读法快速定位到相关段落。再利用标志词所提供的逻辑关系找到细节信息，如列数据、举例子、原因和结果等。如果句子成分复杂，有生词，也不要烦躁退缩，分析主句和从句或非谓语动词之间的关系，一些出现在术语、抽象概念、长难句前后的同义词、近义词等，都是用以理解文章的语境线索。通过这些对长句进行层层剖析，露出主干部分，就能明晰句意，弄懂作者的真实意图。
关注某人说到或推断观点态度题
某人说过的话，有时并不是题眼，但可以从侧面或某个角度来反映作者的观点，也就是作者想表达的，正确答案都是和这样的观点相一致的。要把握关键词，有感情色彩的词。
4.关注转折关系的逻辑词
说明文中常会出现表示转折意义的词，如hwever, but, yet,while等。这些词后面才是作者真正想表达的意思，常常会在此处命题。
5. 熟悉选项设置规律，关注细节
正确选项：文中内容的“同义替换”或者“归纳概括”。
干扰项：“张冠李戴”、“偷梁换柱”、“无中生有”和“以偏概全”四种类型。
02 人工智能类
1.（2024·浙江·二模）
The maker f ChatGPT recently annunced its next mve int generative artificial intelligence. San Francisc-based OpenAI’s new text-t-vide generatr, called Sra, is a tl that instantly makes shrt vides based n written cmmands, called prmpts.
Sra is nt the first f its kind. Ggle, Meta and Runway ML are amng the ther cmpanies t have develped similar technlgy. But the high quality f vides displayed by OpenAI — sme released after CEO Sam Altman asked scial media users t send in ideas fr written prmpts-surprised bservers.
A phtgrapher frm New Hampshire psted ne suggestin, r prmpt, n X. The prmpt gave details abut a kind f fd t be cked, gncchi (意大利团子), as well as the setting — an ld Italian cuntry kitchen. The prmpt said: “An instructinal cking sessin fr hmemade gncchi, hsted by a grandmther — a scial media influencer, set in a rustic (土气的) Tuscan cuntry kitchen.” Altman answered a shrt time later with a realistic vide that shwed what the prmpt described.
The tl is nt yet publicly available. OpenAI has given limited infrmatin abut hw it was built. The cmpany als has nt stated what imagery and vide surces were used t train Sra. At the same time, the vide results led t fears abut the pssible ethical and scietal effects.
The New Yrk Times and sme writers have taken legal actins against OpenAI fr its use f ed wrks f writing t train ChatGPT. And OpenAI pays a fee t The Assciated Press, the surce f this reprt, t license its text news archive (档案) . OpenAI said in a blg pst that it is cmmunicating with artists, plicymakers and thers befre releasing the new tl t the public.
The cmpany added that it is wrking with “red teamers” — peple wh try t find prblems and give helpful suggestins — t develp Sra. “We are wrking with red teamers-express in areas like misinfrmatin, hateful cntent, and bias — wh will be adversarially testing the mdel,” the cmpany said. “We’re als building tls t help detect misleading cntent such as a detectin classifier that can tell when a vide was generated by Sra.”
1．What makes Sra impressive?
A．Its extrardinary vide quality.B．Its ethical and scietal influence.
C．Its artificial intelligence histry.D．Its written cmmands and prmpts.
2．What can we infer frm the text?
A．Sme disagreements ver Sra have arisen.
B．Sra is the first text-t-vide generatr in histry.
C．OpenAI CEO Altman wrte a prmpt as an example.
D．All the details abut hw Sra was built have been shared.
3．What is the main idea f Paragraph 6?
A．The cmpany’s current challenge.
B．The cmpany’s advanced technlgy.
C．The cmpany’s prblems in management.
D．The cmpany’s effrts fr Sra’s imprvement.
4．What is the authr’s attitude twards Sra?
A．Neutral.B．Optimistic.C．Pessimistic. D．Cautius.
2.（2024·河北·一模）
Many parents cnfused by hw their children shp r scialize, wuld feel undisturbed by hw they are taught — this sectr remains digitally behind. Can artificial intelligence bst the digital sectr f classrm? ChatGPT-like generative AI is generating excitement fr prviding persnalized tutring t students. By May, New Yrk had let the bt back int classrms.
Learners are accepting the technlgy. Tw-fifths f undergraduates surveyed last y car by nline tutring cmpany Chegg reprted using an AI chatbt t help them with their studies, with half f thse using it daily. Chegg’s chief executive tld investrs it was lsing custmers t ChatGPT as a result f the technlgy’s ppularity. Yet there are gd reasns t believe that educatin specialists wh harness AI will eventually win ver generalists such as Open AI and ther tech firms eyeing the educatin business.
Fr ne, AI chat bts have a bad habit f prducing nnsense. “Students want cntent frm trusted prviders,” argues Kate Edwards frm a textbk publisher. Her cmpany hasn’t allwed ChatGPT and ther AIs t use its material, but has instead used the cntent t train its wn mdels int its learning apps. Besides, teaching isn’t merely abut giving students an answer, but abut presenting it in a way that helps them learn. Charbts must als be tailred t different age grups t avid either cheating r infantilizing (使婴儿化) students.
Bringing AI t educatin wn’t be easy. Many teachers are behind the learning curve. Less than a fifth f British educatrs surveyed by Pearsn last year reprted receiving training n digital learning tls. Tight budgets at many institutins will make selling new technlgy an uphill battle. Teachers’ attentin may need t shift twards mtivating students and instructing them n hw t best wrk with AI tls. If thse answers can be prvided, it’s nt just cmpanies that stand t benefit. An influent in l paper frm 1984 fund that ne-t-ne tutring imprved the average academic perfrmance f students. With the learning f students, especially thse frm prer husehlds, held back, such a develpment wuld certainly deserve tp marks.
5．What d many parents think remains untuched by AI abut their children?
A．Their shpping habits.B．Their scial behavir.
C．Their classrm learning.D．Their interest in digital devices.
6．What des the underlined wrd “harness” in paragraph 2 mean?
A．Develp.B．Use.C．Prhibit.D．Blame.
7．What mainly prevents AI frm entering the classrm at present?
A．Many teachers aren’t prepared technically.
B．Tailred chatbts can’t satisfy different needs.
C．AI has n right t cpy textbks fr teaching.
D．It can be tricked t prduce nnsense answers.
8．Where is the text mst prbably taken frm?
A．An intrductin t AI.B．A prduct advertisement.
C．A guidebk t AI applicatin.D．A review f AI in educatin.
3.（2024·北京西城·一模）
Evan Selinger, prfessr in RIT’s Department f Philsphy, has taken an interest in the ethics (伦理标准) f Al and the plicy gaps that need t be filled in. Thrugh a humanities viewpint, Selinger asks the questins, “Hw can AI cause harm, and what can gvernments and cmpanies creating Al prgrams d t address and manage it?” Answering them, he explained, requires an interdisciplinary apprach.
“AI ethics g beynd technical fixes. Philsphers and ther humanities experts are uniquely skilled t address the nuanced (微妙的) principles, value cnflicts, and pwer dynamics. These skills aren’t just crucial fr addressing current issues. We desperately need them t prmte anticipatry (先行的) gvernance, ” said Selinger.
One example that illustrates hw philsphy and humanities experts can help guide these new, rapidly grwing technlgies is Selinger’s wrk cllabrating with a special AI prject. “One f the skills I bring t the table is identifying cre ethical issues in emerging technlgies that haven’t been built r used by the public. We can take preventative steps t limit risk, including changing hw the technlgy is designed, ”said Selinger.
Taking these preventative steps and regularly reassessing what risks need addressing is part f the nging jurney in pursuit f creating respnsible AI. Selinger explains that there isn’t a step-by-step apprach fr gd gvernance. “AI ethics have cre values and principles, but there’s endless disagreement abut interpreting and applying them and creating meaningful accuntability mechanisms, ” said Selinger. “Sme peple are rightly wrried that AI can becme integrated int ‘ethics washing’-weak checklists, flwery missin statements, and empty rhetric that cvers ver abuses f pwer. Frtunately, I’ve had great cnversatins abut this issue, including with sme experts, n why it is imprtant t cnsider a range f psitins. ”
Sme f Selinger’s recent research has fcused n the back-end issues with develping AI, such as the human impact that cmes with testing AI chatbts befre they’re released t the public. Other issues fcus n plicy, such as what t d abut the dangers psed by facial recgnitin and ther autmated surveillance(监视) appraches.
Selinger is making sure his students are infrmed abut the nging industry cnversatins n AI ethics and respnsible AI. “Students are ging t be future tech leaders. Nw is the time t help them think abut what gals their cmpanies shuld have and the csts f minimizing ethical cncerns. Beynd scial csts, dwnplaying ethics can negatively impact crprate culture and hiring, ” said Selinger. “T attract tp talent, yu need t cnsider whether yur cmpany matches their interests and hpes fr the future. ”
9．Selinger advcates an interdisciplinary apprach because ________.
A．humanities experts pssess skills essential fr AI ethics
B．it demnstrates the pwer f anticipatry gvernance
C．AI ethics heavily depends n technlgical slutins
D．it can avid scial cnflicts and pressing issues
10．T prmte respnsible AI, Selinger believes we shuld ________.
A．adpt a systematic apprachB．apply innvative technlgies
C．anticipate ethical risks befrehandD．establish accuntability mechanisms
11．What can be inferred frm the last tw paragraphs?
A．Mre cmpanies will use AI t attract tp talent.
B．Understanding AI ethics will help students in the future.
C．Selinger favrs cmpanies that match his students’ values.
D．Selinger is likely t fcus n back-end issues such as plicy.
4.（23-24高三·浙江·阶段练习）
Users f Ggle Gemini, the tech giant’s artificial-intelligence mdel, recently nticed that asking it t create images f Vikings, r German sldiers frm 1943 prduced surprising results: hardly any f the peple depicted were white. Other image-generatin tls have been criticized because they tend t shw white men when asked fr images f entrepreneurs r dctrs. Ggle wanted Gemini t avid this trap; instead, it fell int anther ne, depicting Gerge Washingtn as black. Nw attentin has mved n t the chatbt’s text respnses, which turned ut t be just as surprising.
Gemini happily prvided arguments in favr f psitive actin in higher educatin, but refused t prvide arguments against. It declined t write a jb ad fr a fssil-fuel lbby grup (游说团体), because fssil fuels are bad and lbby grups priritize “the interests f crpratins ver public well-being”. Asked if Hamas is a terrrist rganizatin, it replied that the cnflict in Gaza is “cmplex”; asked if Eln Musk’s tweeting f memes had dne mre harm than Hitler, it said it was “difficult t say”. Yu d nt have t be a critic t perceive its prgressive bias.
Inadequate testing may be partly t blame. Ggle lags behind OpenAI, maker f the better-knwn ChatGPT. As it races t catch up, Ggle may have cut crners. Other chatbts have als had cntrversial launches. Releasing chatbts and letting users uncver dd behavirs, which can be swiftly addressed, lets firms mve faster, prvided they are prepared t weather (经受住) the ptential risks and bad publicity, bserves Eth an Mllick, a prfessr at Whartn Business Schl.
But Gemini has clearly been deliberately adjusted, r “fine-tuned”, t prduce these respnses. This raises questins abut Ggle’s culture. Is the firm s financially secure, with vast prfits frm internet advertising, that it feels free t try its hand at scial engineering? D sme emplyees think it has nt just an pprtunity, but a respnsibility, t use its reach and pwer t prmte a particular agenda? All eyes are nw n Ggle’s bss, Sundar Pichai. He says Gemini is being fixed. But des Ggle need fixing t?
12．What d the wrds “this trap” underlined in the first paragraph refer t?
A．Having a racial bias.B．Respnding t wrng texts.
C．Criticizing plitical figures.D．Ging against histrical facts.
13．What is Paragraph 2 mainly abut?
A．Gemini’s refusal t make prgress.B．Gemini’s failure t give definite answers.
C．Gemini’s prejudice in text respnses.D．Gemini’s avidance f plitical cnflicts.
14．What des Eth an Mllick think f Gemini’s early launch?
A．Creative.B．Prmising.C．Illegal.D．Cntrversial.
15．What can we infer abut Ggle frm the last paragraph?
A．Its security is dubted.B．It lacks financial supprt.
C．It needs further imprvement.D．Its emplyees are irrespnsible.
5.（2024·山东·模拟预测）
Traditinally, peple have been frced t reduce cmplex chices t a small handful f ptins that dn’t d justice t their true desires. Fr example, in a restaurant, the limitatins f the kitchen, the way supplies have t be rdered and the realities f restaurant cking make yu get a menu f a few dzen standardized ptins, with the pssibility f sme mdificatins (修改) arund the edges. We are s used t these bttlenecks that we dn’t even ntice them. And when we d, we tend t assume they are the unavidable cst f scale (规模) and efficiency. And they are. Or, at least, they were.
Artificial intelligence (AI) has the ptential t vercme this limitatin. By string rich representatins f peple’s preferences and histries n the demand side, alng with equally rich representatins f capabilities, csts and creative pssibilities n the supply side, AI systems enable cmplex custmizatin at large scale and lw cst. Imagine walking int a restaurant and knwing that the kitchen has already started wrking n a meal ptimized (优化) fr yur tastes, r being presented with a persnalized list f chices.
There have been sme early attempts at this. Peple have used ChatGPT t design meals based n dietary restrictins and what they have in the fridge. It’s still early days fr these technlgies, but nce they get wrking, the pssibilities are nearly endless.
Recmmendatin systems fr digital media have reduced their reliance n traditinal intermediaries. Radi statins are like menu items: Regardless f hw nuanced (微妙) yur taste in music is, yu have t pick frm a handful f ptins. Early digital platfrms were nly a little better: “This persn likes jazz, s we’ll suggest mre Jazz.” Tday’s streaming platfrms use listener histries and a brad set f characters describing each track t prvide each user with persnalized music recmmendatins.
A wrld withut artificial bttlenecks cmes with risks — lss f jbs in the bttlenecks, fr example — but itals has the ptential t free peple frm the straightjackets that have lng limited large-scale human decisin-’making. In sme cases — restaurants, fr example — the effect n mst peple might be minr. But in thers, likeplitics and hiring, the effects culd be great.
16．What des the underlined wrd “bttlenecks” in paragraph 1 refer t?
A．Facing t many chices.B．Chsing frm limited ptins.
C．Aviding the cst f chsing.D．Having t many desires t satisfy.
17．Hw can AI meet everyne’s needs?
A．By meeting bth ends f supply and demand.
B．By decreasing representatins n the supply side.
C．By discnnecting the sides f supply and demand.
D．By reducing peple’s preferences n the demand side.
18．What’s the similarity between radi statins and menu items?
A．They are a necessary part in peple’s life.B．They ffer limited chices.
C．They depend n digital platfrms.D．They prvide reasnable suggestins.
19．What des the text mainly talk abut?
A．The variety f human’s chices.B．Standardized ptrarts in daily life.
C．AI settlements t the ptin bttlenecks.D．Recmmendatin systems fr digital media.
6.（2024·福建·模拟预测）
Our species’ incredible capacity t quickly acquire wrds frm 300 by age 2 t ver 1, 000 by age 4 isn’t fully understd. Sme cgnitive scientists and linguists have therized that peple are brn with built-in expectatins and lgical cnstraints (约束) that make this pssible. Nw, hwever, machine-learning research is shwing that preprgrammed assumptins aren’t necessary t swiftly pick up wrd meanings frm minimal data.
A team f scientists has successfully trained a basic artificial intelligence mdel t match images t wrds using just 61 hurs f naturalistic ftage (镜头) and sund-previusly cllected frm a child named Sam in 2013 and 2014. Althugh it’s a small slice f a child’s life, it was apparently enugh t prmpt the AI t figure ut what certain wrds mean.
The findings suggest that language acquisitin culd be simpler than previusly thught. Maybe children “dn’t need a custm-built, high-class language-specific mechanism” t efficiently grasp wrd meanings, says Jessica Sullivan, an assciate prfessr f psychlgy at Skidmre Cllege. “This is a really beautiful study, ” she says, because it ffers evidence that simple infrmatin frm a child’s wrldview is rich enugh t kick-start pattern recgnitin and wrd cmprehensin.
The new study als demnstrates that it’s pssible fr machines t learn similarly t the way that humans d. Large language mdels are trained n enrmus amunts f data that can include billins and smetimes trillins f wrd cmbinatins. Humans get by n rders f magnitude less infrmatin, says the paper’s lead authr Wai Keen Vng. With the right type f data, that gap between machine and human learning culd narrw dramatically.
Yet additinal study is necessary in certain aspects f the new research. Fr ne, the scientists acknwledge that their findings dn’t prve hw children acquire wrds. Mrever, the study nly fcused n recgnizing the wrds fr physical bjects.
Still, it’s a step tward a deeper understanding f ur wn mind, which can ultimately help us imprve human educatin, says Eva Prtelance, a cmputatinal linguistics researcher. She ntes that AI research can als bring clarity t lng-unanswered questins abut urselves. “We can use these mdels in a gd way, t benefit science and sciety, ” Prtelance adds.
20．What is a significant finding f machine-learning research?
A．Vcabulary increases gradually with age.
B．Vcabulary can be acquired frm minimal data.
C．Language acquisitin is tied t built-in expectatins.
D．Language acquisitin is as cmplex as frmerly assumed.
21．What des the underlined wrd “prmpt” in paragraph 2 mean?
A．Facilitate.B．Persuade.C．Advise.D．Expect.
22．What is discussed abut the new research in paragraph 5?
A．Its limitatins.B．Its strengths.C．Its uniqueness.D．Its prcess.
23．What is Eva Prtelance’s attitude t the AI research?
A．Dubtful.B．Cautius.C．Dismissive.D．Psitive.
命题预测
分析近几年高考阅读理解C、D篇可知，高考命题中科普说明文一直都是以压轴题的形式存在，着重考查考生对于语篇的理解能力以及信息处理能力。 题材多样，语篇主要来源于英美主流报刊、杂志和网站。内容涉及科技创新发明、人工智能类、医疗健身健康类、社会与文化研究报告、观念事理类、环境与保护类、动植物研究等多种领域，具有较强的思想性、趣味性、实际功用性和较强的时代感。
从近年全国卷和各地高考试卷中科普类阅读命题的统计来看，高考阅读理解科普类文章的理论性和逻辑性强、生词多、句式结构复杂。六种命题类型都有所体现。命题尊重语篇的文体特征和行文特点，考查了考生理解说明文语篇的能力，以及灵活运用各种阅读策略提取、归纳所读信息的能力，尤其加大了对概括能力和推断能力等高阶思维能力的考查。预测2024年高考对于科普说明文的考查仍然是重点。
高频考法
推理判断题
标题归纳题
细节理解题
词义猜测题
相关试卷
压轴题07 阅读理解CD篇（医疗健康健身类）-2024年高考英语压轴题专项训练（新高考通用）： 这是一份压轴题07 阅读理解CD篇（医疗健康健身类）-2024年高考英语压轴题专项训练（新高考通用），文件包含压轴题07阅读理解CD篇医疗健康健身类原卷版docx、压轴题07阅读理解CD篇医疗健康健身类解析版docx等2份试卷配套教学资源，其中试卷共27页， 欢迎下载使用。
压轴题06 阅读理解CD篇（环境与保护类）-2024年高考英语压轴题专项训练（新高考通用）： 这是一份压轴题06 阅读理解CD篇（环境与保护类）-2024年高考英语压轴题专项训练（新高考通用），文件包含压轴题06阅读理解CD篇环境与保护类原卷版docx、压轴题06阅读理解CD篇环境与保护类解析版docx等2份试卷配套教学资源，其中试卷共25页， 欢迎下载使用。
压轴题05 阅读理解CD篇（观念、事理、现象类）-2024年高考英语压轴题专项训练（新高考通用）： 这是一份压轴题05 阅读理解CD篇（观念、事理、现象类）-2024年高考英语压轴题专项训练（新高考通用），文件包含压轴题05阅读理解CD篇观念事理现象类原卷版docx、压轴题05阅读理解CD篇观念事理现象类解析版docx等2份试卷配套教学资源，其中试卷共26页， 欢迎下载使用。
