登录
D
新素养 批判质疑 Can a machine think? Can a computer answer questions in a human-like way? How can you tell if you are talking to a computer?
New computer programmes called language models have been very good at copying people. These artificial intelligence(AI) programmes analyze(分析) millions of sentences from books and websites to learn hidden patterns in language. It's exciting to see AI making realistic sentences. But there are also problems with AI-made texts. People have used AI-made texts to cheat on tests.
We wanted to learn more about AI. So we called up 4,600 participants(参与者) for a study. We told the participants that they were using a website where some people wrote their own biographies(传记), and some people used AI to write theirs. Each participant rated(评分) 16 texts, half of which were AI-made. They then rated each text on a 5-point scale(评分等级) from "definitely(确定) AI" to "definitely human". The participants were largely unable to tell whether they were people-made or AI-made. Their answers were wrong about 49% of the time.
We found that people were following a few rules of thumb(经验法则). When people saw a meaningless or repetitive(重复的) text, they were more likely to say it came from AI. Truly, an AI text was more likely to have these qualities. But some rules of thumb gave wrong answers! People thought bad grammar and unusual word choices appear in an AI text, but they were actually more likely to come from human-made texts. And when people saw contractions (like "don't" or "won't") they thought it was a sign of a human writer, but actually, AI was more likely to use such words.
These are often wrong, which means that people are likely to be cheated. A person sends emails to people pretending(假装) to be someone you believe in. The email might say "Hi, I work for your bank! You need to log in and fix something in your account." With AI, the person could make different emails to try to pretend one that's very believable.
There is a lot of useful information online, and a lot of fun things too. But there are also a lot of ways to get in trouble. We can never know who is on the other side when talking to someone on the internet—it may not even be a human! That's why it's important not to just rely on your feeling, but check the website URL and author information. If something seems strange, tell your parents or a responsible adult about it!
(
A. who can write biographies faster, human or AI
B. whether AI can answer questions in a human-like way
C. which biographies were better, the human-made or the AI-made
D. whether we can tell the human-made biographies from the AI-made
(
A. Answers.
B. Websites.
C. Biographies.
D. Participants.
(
A. Engineers are developing AI's copying skills.
B. The participants use a website to write their biographies.
C. There are more wrong sentence structures in AI-made texts.
D. AI-made texts are likely to choose the same words to express.
(
A. To compare(比较) AI's writing with human's.
B. To study the reasons of popularity of AI use.
C. To improve our judgement(判断) online in the age of AI.
D. To discuss the main characteristics(特征) of AI-made writing.
(
A. AI language models will replace human writers in the future
B. people are becoming more and more skilled at using AI
C. AI-made texts are completely the same as human-made ones
D. being careful is necessary when dealing with online information
新素养 批判质疑 Can a machine think? Can a computer answer questions in a human-like way? How can you tell if you are talking to a computer?
New computer programmes called language models have been very good at copying people. These artificial intelligence(AI) programmes analyze(分析) millions of sentences from books and websites to learn hidden patterns in language. It's exciting to see AI making realistic sentences. But there are also problems with AI-made texts. People have used AI-made texts to cheat on tests.
We wanted to learn more about AI. So we called up 4,600 participants(参与者) for a study. We told the participants that they were using a website where some people wrote their own biographies(传记), and some people used AI to write theirs. Each participant rated(评分) 16 texts, half of which were AI-made. They then rated each text on a 5-point scale(评分等级) from "definitely(确定) AI" to "definitely human". The participants were largely unable to tell whether they were people-made or AI-made. Their answers were wrong about 49% of the time.
We found that people were following a few rules of thumb(经验法则). When people saw a meaningless or repetitive(重复的) text, they were more likely to say it came from AI. Truly, an AI text was more likely to have these qualities. But some rules of thumb gave wrong answers! People thought bad grammar and unusual word choices appear in an AI text, but they were actually more likely to come from human-made texts. And when people saw contractions (like "don't" or "won't") they thought it was a sign of a human writer, but actually, AI was more likely to use such words.
These are often wrong, which means that people are likely to be cheated. A person sends emails to people pretending(假装) to be someone you believe in. The email might say "Hi, I work for your bank! You need to log in and fix something in your account." With AI, the person could make different emails to try to pretend one that's very believable.
There is a lot of useful information online, and a lot of fun things too. But there are also a lot of ways to get in trouble. We can never know who is on the other side when talking to someone on the internet—it may not even be a human! That's why it's important not to just rely on your feeling, but check the website URL and author information. If something seems strange, tell your parents or a responsible adult about it!
(
D
)41. The researchers did the study to find out .A. who can write biographies faster, human or AI
B. whether AI can answer questions in a human-like way
C. which biographies were better, the human-made or the AI-made
D. whether we can tell the human-made biographies from the AI-made
(
C
)42. What does the underlined word "they" in Paragraph 3 refer to?A. Answers.
B. Websites.
C. Biographies.
D. Participants.
(
D
)43. Which of the following statements is TRUE according to the passage?A. Engineers are developing AI's copying skills.
B. The participants use a website to write their biographies.
C. There are more wrong sentence structures in AI-made texts.
D. AI-made texts are likely to choose the same words to express.
(
C
)44. 新考法 写作意图 What is the writer's main purpose in writing the passage?A. To compare(比较) AI's writing with human's.
B. To study the reasons of popularity of AI use.
C. To improve our judgement(判断) online in the age of AI.
D. To discuss the main characteristics(特征) of AI-made writing.
(
D
)45. From the passage, we can infer that .A. AI language models will replace human writers in the future
B. people are becoming more and more skilled at using AI
C. AI-made texts are completely the same as human-made ones
D. being careful is necessary when dealing with online information
答案
41. D 42. C 43. D 44. C 45. D
解析
【分析】
1. 第41题：解题思路是定位研究目的。原文第二段明确提到研究召集参与者对文本（一半AI生成、一半人类创作）评分并判断来源，核心是探究能否区分人类与AI创作的文本，对应选项D。
2. 第42题：解题思路是代词指代，需结合上下文语境。第三段前文提到“人类写传记、AI也写传记，参与者给16篇文本评分”，此处“they”指代被评分的传记，对应选项C。
3. 第43题：解题思路是细节判断，逐一核对选项。A项原文未提及工程师开发AI复制技能；B项参与者是评分而非写传记；C项原文指出人类文本更可能有糟糕语法；D项原文提到AI更可能使用缩写词，符合选项描述，故D正确。
4. 第44题：解题思路是归纳写作意图。原文最后一段强调“不要仅靠感觉判断网络信息，要检查来源、遇到异常找成人”，核心是提高AI时代的网络判断能力，对应选项C。
5. 第45题：解题思路是合理推断。最后一段提到“网络另一端可能不是人类，需小心”，对应选项D；A项未提AI取代人类作家，B项未提人们更熟练用AI，C项“完全相同”过于绝对，均排除。
【解析】
41题：根据原文第二段研究过程（召集参与者对AI和人类创作的文本评分并判断来源），可知研究目的是区分人类与AI创作的文本，选D。
42题：第三段中，前文提及“人类和AI都写传记，参与者给16篇文本评分”，此处“they”指代被评分的传记，选C。
43题：A项原文未涉及；B项参与者是评分而非写传记；C项原文说明人类文本更可能有糟糕语法；D项原文提到AI更可能使用缩写词，符合描述，选D。
44题：原文最后一段强调需谨慎判断网络信息，核心是提高AI时代的网络判断能力，选C。
45题：A项无依据；B项未提及；C项“完全相同”表述绝对；D项符合最后一段“要小心网络信息”的推断，选D。
【答案】
41. D 42. C 43. D 44. C 45. D
【知识点】
阅读理解、细节理解、主旨大意
【点评】
本题为英语阅读理解题，围绕AI文本与人类文本的区分展开，考查学生对原文细节的查找、代词指代的理解、主旨的归纳及合理推断能力，需结合原文信息逐一分析选项。
【难度系数】
0.6
1. 第41题：解题思路是定位研究目的。原文第二段明确提到研究召集参与者对文本（一半AI生成、一半人类创作）评分并判断来源，核心是探究能否区分人类与AI创作的文本，对应选项D。
2. 第42题：解题思路是代词指代，需结合上下文语境。第三段前文提到“人类写传记、AI也写传记，参与者给16篇文本评分”，此处“they”指代被评分的传记，对应选项C。
3. 第43题：解题思路是细节判断，逐一核对选项。A项原文未提及工程师开发AI复制技能；B项参与者是评分而非写传记；C项原文指出人类文本更可能有糟糕语法；D项原文提到AI更可能使用缩写词，符合选项描述，故D正确。
4. 第44题：解题思路是归纳写作意图。原文最后一段强调“不要仅靠感觉判断网络信息，要检查来源、遇到异常找成人”，核心是提高AI时代的网络判断能力，对应选项C。
5. 第45题：解题思路是合理推断。最后一段提到“网络另一端可能不是人类，需小心”，对应选项D；A项未提AI取代人类作家，B项未提人们更熟练用AI，C项“完全相同”过于绝对，均排除。
【解析】
41题：根据原文第二段研究过程（召集参与者对AI和人类创作的文本评分并判断来源），可知研究目的是区分人类与AI创作的文本，选D。
42题：第三段中，前文提及“人类和AI都写传记，参与者给16篇文本评分”，此处“they”指代被评分的传记，选C。
43题：A项原文未涉及；B项参与者是评分而非写传记；C项原文说明人类文本更可能有糟糕语法；D项原文提到AI更可能使用缩写词，符合描述，选D。
44题：原文最后一段强调需谨慎判断网络信息，核心是提高AI时代的网络判断能力，选C。
45题：A项无依据；B项未提及；C项“完全相同”表述绝对；D项符合最后一段“要小心网络信息”的推断，选D。
【答案】
41. D 42. C 43. D 44. C 45. D
【知识点】
阅读理解、细节理解、主旨大意
【点评】
本题为英语阅读理解题，围绕AI文本与人类文本的区分展开，考查学生对原文细节的查找、代词指代的理解、主旨的归纳及合理推断能力，需结合原文信息逐一分析选项。
【难度系数】
0.6
