hce_cmu
115年
英文
第 39 題
📖 題組：
In contemporary higher education, the discourse surrounding generative artificial intelligence (AI) has largely coalesced around a set of comforting metaphors. University policies and academic handbooks frequently characterize AI as a “tool,” a “tutor,” or a “helpful assistant.” While these descriptors aim to provide pragmatic clarity, they are far from neutral. Instead, these metaphors exert a profound, often invisible, influence on how educators and students conceptualize the technology’s role in the cultivation of critical thinking. The cognitive power of metaphors is well-documented. A landmark study by Stanford psychologists Paul Thibodeau and Lera Boroditsky demonstrated that framing crime as either a “beast” or a “plague” significantly altered the solutions participants proposed—steering them toward either punitive measures or social reform, respectively. Crucially, participants remained largely unaware of the metaphor’s influence, instead attributing their decisions to raw statistics. Similarly, viewing AI as a “tool” implies a degree of moral neutrality and human agency; much like a hammer, the responsibility for its effects is placed squarely on the user. This perspective, however, obscures the ways in which AI systems actively shape interpretations and manipulate judgment, potentially undermining the intellectual habits academics seek to foster. The “assistant” metaphor further complicates the pedagogical landscape by suggesting a clear hierarchy where the human remains in control. This narrative masks the emergence of a “second hidden curriculum,” wherein AI does not merely assist but actively directs learning by structuring explanations and modeling specific cognitive pathways. When AI is anthropomorphized—attributed human qualities like “going rogue” or being “racist”—it further dilutes accountability. Such language allows users to look away from the institutional and corporate biases embedded within the algorithms, leading to a drift in moral responsibility. To counteract these cognitive distortions, some scholars suggest a transition from convenient metaphors to disciplined, technical vocabulary. Rather than “brainstorming with AI,” users should recognize the process as “engaging in probabilistic text generation” under specific algorithmic constraints. Shifting from the concept of “hallucinations” to “predictive text failure” transforms verification from an optional task into a fundamental academic practice. Ultimately, by stripping away misleading metaphors and insisting on technical precision, the academic community can better maintain moral responsibility and pedagogical integrity, ensuring that the development of thought remains a distinctly human endeavor.
In contemporary higher education, the discourse surrounding generative artificial intelligence (AI) has largely coalesced around a set of comforting metaphors. University policies and academic handbooks frequently characterize AI as a “tool,” a “tutor,” or a “helpful assistant.” While these descriptors aim to provide pragmatic clarity, they are far from neutral. Instead, these metaphors exert a profound, often invisible, influence on how educators and students conceptualize the technology’s role in the cultivation of critical thinking. The cognitive power of metaphors is well-documented. A landmark study by Stanford psychologists Paul Thibodeau and Lera Boroditsky demonstrated that framing crime as either a “beast” or a “plague” significantly altered the solutions participants proposed—steering them toward either punitive measures or social reform, respectively. Crucially, participants remained largely unaware of the metaphor’s influence, instead attributing their decisions to raw statistics. Similarly, viewing AI as a “tool” implies a degree of moral neutrality and human agency; much like a hammer, the responsibility for its effects is placed squarely on the user. This perspective, however, obscures the ways in which AI systems actively shape interpretations and manipulate judgment, potentially undermining the intellectual habits academics seek to foster. The “assistant” metaphor further complicates the pedagogical landscape by suggesting a clear hierarchy where the human remains in control. This narrative masks the emergence of a “second hidden curriculum,” wherein AI does not merely assist but actively directs learning by structuring explanations and modeling specific cognitive pathways. When AI is anthropomorphized—attributed human qualities like “going rogue” or being “racist”—it further dilutes accountability. Such language allows users to look away from the institutional and corporate biases embedded within the algorithms, leading to a drift in moral responsibility. To counteract these cognitive distortions, some scholars suggest a transition from convenient metaphors to disciplined, technical vocabulary. Rather than “brainstorming with AI,” users should recognize the process as “engaging in probabilistic text generation” under specific algorithmic constraints. Shifting from the concept of “hallucinations” to “predictive text failure” transforms verification from an optional task into a fundamental academic practice. Ultimately, by stripping away misleading metaphors and insisting on technical precision, the academic community can better maintain moral responsibility and pedagogical integrity, ensuring that the development of thought remains a distinctly human endeavor.
Which of the following is NOT mentioned as a consequence of anthropomorphizing AI?
- A It makes it easier to ignore corporate choices embedded in the AI system.
- B It leads to the idea that the AI system has “gone rogue” when it produces errors.
- C It encourages users to think using the AI system is a key to success.
- D It allows users to take an evasive attitude in moral responsibility.
思路引導 VIP
請觀察文章第三段，當我們賦予 AI 人類特質（例如形容它「失控」或「有偏見」）時，作者認為這會如何影響人們對於「責任歸屬」以及「演算法背後的公司決策」的看法？請試著找出描述這些影響的關鍵句。
🤖
AI 詳解
AI 專屬家教
太棒了！你非常精準地捕捉到了文章中的細節。這類「否定型提問」（NOT mentioned）最考驗閱讀的耐心與細緻度，而你成功避開了干擾選項。
擬人化修辭的負面代價
文中第三段深入探討了將 AI 擬人化（Anthropomorphizing） 所帶來的連鎖反應。當我們賦予 AI 如「失控（going rogue）」或「種族歧視」等人類特質時，會產生一種遮蔽效應，讓使用者忽視了埋藏在演算法背後的企業決策與體制偏見（對應選項 A），進而導致道德責任的稀釋與逃避（對應選項 D）。選項 B 則直接對應了文中提到的「going rogue」具體描述。相對地，選項 C 提到的「AI 是成功的關鍵」雖然在一般廣告中常見，但在本文討論擬人化如何導致「責任流失」的邏輯脈絡中，完全沒有被提及。
▼ 還有更多解析內容