๐Ÿค–

The Ethics of Artificial Intelligence

Agent Progress
Built
Keywords
์ธ๊ณต์ง€๋Šฅ ์œค๋ฆฌ / ์ฑ…์ž„ / ํŽธํ–ฅ ๋ฐฉ์ง€ / ์ธ๊ฐ„์„ฑ ๋ณดํ˜ธ
Lesson #
S
Done
Section
Advanced Vocabulary
๐ŸŽฏ
TL;DR ๐Ÿฆ„
  • ๐Ÿค– AI ethics is the vocabulary family for asking, โ€œShould this system do that, and who is responsible if it harms someone?โ€
  • โš–๏ธ The core terms are ์ธ๊ณต์ง€๋Šฅ ์œค๋ฆฌ (AI ethics), ์ฑ…์ž„ (responsibility), ํŽธํ–ฅ ๋ฐฉ์ง€ (bias prevention), and ์ธ๊ฐ„์„ฑ ๋ณดํ˜ธ (protection of humanity).
  • ๐Ÿงญ Advanced Korean discussions often pair these nouns with verbs like ๋ณด์žฅํ•˜๋‹ค (to guarantee), ์˜ˆ๋ฐฉํ•˜๋‹ค (to prevent), ๊ทœ์ œํ•˜๋‹ค (to regulate), and ํ›ผ์†ํ•˜๋‹ค (to damage).
๐Ÿ”Š
๐ŸŽง Iru Helper Note โ€” Use
Korean Helper Widgets
Korean Helper Widgets
whenever you want to hear these advanced phrases aloud. Try typing full expressions like ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ํŽธํ–ฅ์„ ๋ฐฉ์ง€ํ•ด์•ผ ํ•œ๋‹ค (algorithmic bias must be prevented) so your ears can catch the rhythm. ๐ŸŽง
ย 

๐ŸŒˆ The Big Picture

Talking about artificial intelligence in Korean is not only about shiny robots and dramatic sci-fi music. It is also about public responsibility, data fairness, human dignity, and institutional accountability. In Korean, this topic usually sounds formal, policy-oriented, and debate-friendly. ๐Ÿ›๏ธ
AI ethics vocabulary often uses compact noun phrases. For example, ์ธ๊ณต์ง€๋Šฅ ์œค๋ฆฌ means โ€œAI ethics,โ€ while ์ฑ…์ž„ ์†Œ์žฌ means โ€œwhere responsibility liesโ€ or โ€œwho is accountable.โ€ These are high-value phrases for essays, presentations, policy discussions, and advanced reading. ๐Ÿ“š
The key move is to connect abstract nouns with action verbs: ํŽธํ–ฅ์„ ๋ฐฉ์ง€ํ•˜๋‹ค (to prevent bias), ํˆฌ๋ช…์„ฑ์„ ํ™•๋ณดํ•˜๋‹ค (to secure transparency), ์ธ๊ฐ„์˜ ์กด์—„์„ฑ์„ ๋ณดํ˜ธํ•˜๋‹ค (to protect human dignity), and ์ฑ…์ž„์„ ๋ช…ํ™•ํžˆ ํ•˜๋‹ค (to clarify responsibility). ๐Ÿงฉ

๐ŸŽจ The Core AI Ethics Family

  • ์ธ๊ณต์ง€๋Šฅ ์œค๋ฆฌ โ†’ AI ethics; the standards and principles used to decide whether AI development and use are socially responsible. ๐Ÿค–
  • ์ฑ…์ž„ โ†’ responsibility; use it for moral, legal, or institutional duty. In AI contexts, ์ฑ…์ž„ ์†Œ์žฌ means โ€œthe locus of responsibility,โ€ or who should be held accountable. โš–๏ธ
  • ํŽธํ–ฅ โ†’ bias; especially unfair distortion in data, algorithms, or judgment. ์•Œ๊ณ ๋ฆฌ์ฆ˜ ํŽธํ–ฅ means โ€œalgorithmic bias.โ€ ๐Ÿงฎ
  • ํŽธํ–ฅ ๋ฐฉ์ง€ โ†’ bias prevention; a policy or design goal that tries to stop unfair outcomes before they spread. ๐Ÿ›ก๏ธ
  • ์ธ๊ฐ„์„ฑ ๋ณดํ˜ธ โ†’ protection of humanity; a broader phrase for preserving human dignity, autonomy, empathy, and moral judgment. ๐Ÿซถ
  • ํˆฌ๋ช…์„ฑ โ†’ transparency; the quality of being explainable, inspectable, or open to review. ๐Ÿ”
  • ๊ณต์ •์„ฑ โ†’ fairness; a key value when AI affects hiring, credit, education, medicine, or public services. ๐Ÿง‘โ€โš–๏ธ
  • ๊ฐœ์ธ์ •๋ณด ๋ณดํ˜ธ โ†’ personal information protection; privacy protection, especially around data collection and use. ๐Ÿ”
  • ๊ทœ์ œ โ†’ regulation; rules or controls set by institutions or governments. ๐Ÿ›๏ธ
  • ๊ฐ๋… โ†’ oversight or supervision; human or institutional monitoring of AI systems. ๐Ÿ‘€
  • ์˜ค๋‚จ์šฉ โ†’ misuse and abuse; a compact noun used for harmful or improper use. ๐Ÿšจ
  • ์ž์œจ์„ฑ โ†’ autonomy; in ethics, the ability of humans to make meaningful choices. ๐Ÿงญ
  • ์กด์—„์„ฑ โ†’ dignity; often appears as ์ธ๊ฐ„์˜ ์กด์—„์„ฑ (human dignity). ๐ŸŒฑ

๐Ÿ› ๏ธ How They Work in a Sentence

  • ์ธ๊ณต์ง€๋Šฅ ์œค๋ฆฌ๋Š” ๊ธฐ์ˆ  ๋ฐœ์ „๊ณผ ์ธ๊ฐ„์˜ ์กด์—„์„ฑ ์‚ฌ์ด์˜ ๊ท ํ˜•์„ ์š”๊ตฌํ•œ๋‹ค. (AI ethics requires balance between technological development and human dignity.) โš–๏ธ
    • Here, ์‚ฌ์ด์˜ ๊ท ํ˜• means โ€œbalance between,โ€ and ์š”๊ตฌํ•˜๋‹ค means โ€œto require.โ€
  • ๊ธฐ์—…์€ ์•Œ๊ณ ๋ฆฌ์ฆ˜ ํŽธํ–ฅ์„ ๋ฐฉ์ง€ํ•˜๊ธฐ ์œ„ํ•ด ํ•™์Šต ๋ฐ์ดํ„ฐ๋ฅผ ์ ๊ฒ€ํ•ด์•ผ ํ•œ๋‹ค. (Companies must inspect training data in order to prevent algorithmic bias.) ๐Ÿงฎ
    • ๋ฐฉ์ง€ํ•˜๊ธฐ ์œ„ํ•ด means โ€œin order to prevent,โ€ and ์ ๊ฒ€ํ•˜๋‹ค means โ€œto inspect/check.โ€
  • ์ž๋™ํ™”๋œ ๊ฒฐ์ •์˜ ์ฑ…์ž„ ์†Œ์žฌ๋ฅผ ๋ช…ํ™•ํžˆ ํ•˜์ง€ ์•Š์œผ๋ฉด ํ”ผํ•ด ๊ตฌ์ œ๊ฐ€ ์–ด๋ ค์›Œ์ง„๋‹ค. (If the locus of responsibility for automated decisions is not clarified, remedying harm becomes difficult.) ๐Ÿงพ
    • ๋ช…ํ™•ํžˆ ํ•˜๋‹ค means โ€œto clarify,โ€ and ํ”ผํ•ด ๊ตฌ์ œ means โ€œremedy/relief for harm.โ€

๐Ÿงฉ General Formation Rule

Advanced AI ethics phrases often form as abstract noun + policy/action noun:
  • ํŽธํ–ฅ (bias) + ๋ฐฉ์ง€ (prevention) โ†’ ํŽธํ–ฅ ๋ฐฉ์ง€ (bias prevention)
  • ๊ฐœ์ธ์ •๋ณด (personal information) + ๋ณดํ˜ธ (protection) โ†’ ๊ฐœ์ธ์ •๋ณด ๋ณดํ˜ธ (privacy protection)
  • ์ฑ…์ž„ (responsibility) + ์†Œ์žฌ (where something lies) โ†’ ์ฑ…์ž„ ์†Œ์žฌ (where responsibility lies)
  • ์ธ๊ฐ„์„ฑ (humanity) + ๋ณดํ˜ธ (protection) โ†’ ์ธ๊ฐ„์„ฑ ๋ณดํ˜ธ (protection of humanity)
These phrases become natural when paired with verbs:
  • ๋ณด์žฅํ•˜๋‹ค โ†’ to guarantee
  • ํ™•๋ณดํ•˜๋‹ค โ†’ to secure
  • ๊ฐ•ํ™”ํ•˜๋‹ค โ†’ to strengthen
  • ํ›ผ์†ํ•˜๋‹ค โ†’ to damage/undermine
  • ์ดˆ๋ž˜ํ•˜๋‹ค โ†’ to cause/bring about
  • ๊ทœ์ œํ•˜๋‹ค โ†’ to regulate

๐Ÿ“š Cross-check note

Key Korean vocabulary and usage patterns were checked against ๊ตญ๋ฆฝ๊ตญ์–ด์› / korean.go.kr, including public corpus and Korean-learning vocabulary resources that list terms such as ์ธ๊ฐ„์„ฑ (humanity) and research materials discussing ์ธ๊ณต์ง€๋Šฅ (artificial intelligence), ํŽธํ–ฅ์„ฑ (bias), and ethical language in Korean corpora. โœ…

๐Ÿชž Synonym Alert

  • ์ธ๊ณต์ง€๋Šฅ / AI โ†’ both mean artificial intelligence; ์ธ๊ณต์ง€๋Šฅ is more Korean and formal, while AI is common in media and business. ๐Ÿค–
  • ํŽธํ–ฅ / ํŽธ๊ฒฌ โ†’ ํŽธํ–ฅ is bias or skew in systems/data/judgment; ํŽธ๊ฒฌ is prejudice, often human/social. ๐Ÿง 
  • ์ฑ…์ž„ / ์ฑ…๋ฌด โ†’ ์ฑ…์ž„ is general responsibility; ์ฑ…๋ฌด is more formal, duty-bound responsibility. ๐Ÿ“œ
  • ๊ฐ๋… / ๊ด€๋ฆฌ โ†’ ๊ฐ๋… emphasizes oversight; ๊ด€๋ฆฌ emphasizes management or administration. ๐Ÿ‘€
  • ๋ณดํ˜ธ / ๋ณด์žฅ โ†’ ๋ณดํ˜ธ means protection from harm; ๋ณด์žฅ means guaranteeing that a right or condition is secured. ๐Ÿ›ก๏ธ

โš ๏ธ Pitfalls to Dodge

  • โŒ AI๊ฐ€ ์ฑ…์ž„์„ ๊ฐ€์ง„๋‹ค โ†’ โœ… AI ๊ฐœ๋ฐœ์ž์™€ ์šด์˜์ž๊ฐ€ ์ฑ…์ž„์„ ์ง„๋‹ค (In Korean ethics discussions, responsibility is usually assigned to people or institutions, not the tool itself.) โš–๏ธ
  • โŒ ํŽธ๊ฒฌ ๋ฐฉ์ง€ for algorithmic fairness โ†’ โœ… ํŽธํ–ฅ ๋ฐฉ์ง€ (Use ํŽธํ–ฅ for data/system bias; ํŽธ๊ฒฌ is more like prejudice.) ๐Ÿงฎ
  • โŒ ์ธ๊ฐ„์„ฑ ๋ณด์žฅ as a default phrase โ†’ โœ… ์ธ๊ฐ„์„ฑ ๋ณดํ˜ธ or ์ธ๊ฐ„์˜ ์กด์—„์„ฑ ๋ณด์žฅ (Protect humanity; guarantee dignity/rights.) ๐ŸŒฑ
  • โŒ ํˆฌ๋ช…ํ•œ AI only โ†’ โœ… ์„ค๋ช… ๊ฐ€๋Šฅํ•œ ์ธ๊ณต์ง€๋Šฅ / ํˆฌ๋ช…ํ•œ ์˜์‚ฌ๊ฒฐ์ • ๊ณผ์ • (For systems, โ€œexplainable AIโ€ or transparent decision-making often sounds more precise.) ๐Ÿ”
  • โŒ ๊ทœ์ œ๋Š” ํ˜์‹ ์„ ๋ฌด์กฐ๊ฑด ๋ง‰๋Š”๋‹ค โ†’ โœ… ์ ์ ˆํ•œ ๊ทœ์ œ๋Š” ํ˜์‹ ๊ณผ ์•ˆ์ „์˜ ๊ท ํ˜•์„ ๋„๋ชจํ•œ๋‹ค (Advanced discussions need balanced framing, not cartoon-villain regulation.) ๐Ÿ›๏ธ

โœจ If you can connect ์ฑ…์ž„, ํŽธํ–ฅ, ํˆฌ๋ช…์„ฑ, and ์กด์—„์„ฑ, you can discuss AI ethics in Korean with nuance instead of just pointing at a robot and whispering โ€œscary.โ€ ๐Ÿค–๐Ÿซถ
ย 

Try this exercise

Practice choosing the strongest advanced Korean term for each AI ethics sentence. Read the full sentence, fill the blank, then open the Answer toggle to check the nuance.
1. ์ธ๊ณต์ง€๋Šฅ ์œค๋ฆฌ๋Š” ์ธ๊ฐ„์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ๋ณดํ˜ธํ•˜๋Š” ๋ฐฉํ–ฅ์œผ๋กœ ๋…ผ์˜๋˜์–ด์•ผ ํ•œ๋‹ค.
AI ethics should be discussed in a direction that protects human dignity. ๐ŸŒฑ
Answer
์กด์—„์„ฑ โ€” ์ธ๊ฐ„์˜ ์กด์—„์„ฑ is the natural formal phrase for human dignity.
2. ์ฑ„์šฉ ์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ์„ฑ๋ณ„์ด๋‚˜ ๋‚˜์ด์— ๋”ฐ๋ฅธ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ์ค„์—ฌ์•ผ ํ•œ๋‹ค.
A hiring algorithm should reduce bias based on gender or age. ๐Ÿง‘โ€๐Ÿ’ผ
Answer
ํŽธํ–ฅ โ€” Use ํŽธํ–ฅ for systemic or data-driven bias.
3. ์ž๋™ํ™”๋œ ๊ฒฐ์ •์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋ฅผ ๋ช…ํ™•ํžˆ ํ•ด์•ผ ํ”ผํ•ด์ž๋ฅผ ๋ณดํ˜ธํ•  ์ˆ˜ ์žˆ๋‹ค.
The locus of responsibility for automated decisions must be clarified to protect victims. ๐Ÿงพ
Answer
์ฑ…์ž„ ์†Œ์žฌ โ€” This asks where responsibility lies.
4. ๊ฐœ์ธ์ •๋ณด ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ์—†์ด๋Š” ์‹ ๋ขฐํ•  ์ˆ˜ ์žˆ๋Š” AI ์„œ๋น„์Šค๋ฅผ ๋งŒ๋“ค๊ธฐ ์–ด๋ ต๋‹ค.
Without personal information protection, it is difficult to create trustworthy AI services. ๐Ÿ”
Answer
๋ณดํ˜ธ โ€” ๊ฐœ์ธ์ •๋ณด ๋ณดํ˜ธ means privacy/personal data protection.
5. ์„ค๋ช… ๊ฐ€๋Šฅํ•œ ์ธ๊ณต์ง€๋Šฅ์€ ์˜์‚ฌ๊ฒฐ์ • ๊ณผ์ •์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ๋†’์ธ๋‹ค.
Explainable AI increases transparency in the decision-making process. ๐Ÿ”
Answer
ํˆฌ๋ช…์„ฑ โ€” ํˆฌ๋ช…์„ฑ is transparency or openness to review.
6. ์ •๋ถ€๋Š” ๊ณ ์œ„ํ—˜ AI์˜ ์˜ค๋‚จ์šฉ์„ ๋ง‰๊ธฐ ์œ„ํ•ด ํ•ฉ๋ฆฌ์ ์ธ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋ฅผ ๋งˆ๋ จํ•ด์•ผ ํ•œ๋‹ค.
The government should prepare reasonable regulation to prevent misuse and abuse of high-risk AI. ๐Ÿ›๏ธ
Answer
๊ทœ์ œ โ€” ๊ทœ์ œ means regulation or control.
7. ์˜๋ฃŒ AI๋Š” ์ง„๋‹จ ๊ฒฐ๊ณผ์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ๋ณด์žฅํ•ด์•ผ ํ•œ๋‹ค.
Medical AI must guarantee fairness in diagnostic results. ๐Ÿฅ
Answer
๊ณต์ •์„ฑ โ€” ๊ณต์ •์„ฑ is fairness.
8. AI ์‹œ์Šคํ…œ์—๋Š” ์ธ๊ฐ„์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์ด ๋ฐ˜๋“œ์‹œ ํ•„์š”ํ•˜๋‹ค.
AI systems absolutely need human oversight. ๐Ÿ‘€
Answer
๊ฐ๋… โ€” ๊ฐ๋… means oversight/supervision.
9. ๋ฌด๋ถ„๋ณ„ํ•œ ์ž๋™ํ™”๋Š” ์ธ๊ฐ„์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ์•ฝํ™”์‹œํ‚ฌ ์ˆ˜ ์žˆ๋‹ค.
Indiscriminate automation may weaken human autonomy. ๐Ÿงญ
Answer
์ž์œจ์„ฑ โ€” ์ž์œจ์„ฑ means autonomy.
10. ๊ธฐ์—…์€ AI ์„œ๋น„์Šค์˜ ์œ„ํ—˜์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•˜์ง€ ๋ง๊ณ  ๊ณต๊ฐœํ•ด์•ผ ํ•œ๋‹ค.
Companies should disclose the risks of AI services rather than concealing them. ๐Ÿšจ
Answer
์€ํํ•˜๋‹ค โ€” ์€ํํ•˜๋‹ค means to conceal; disclosure builds trust.
๐Ÿ“š
Vocabulary Box
  • ์กด์—„์„ฑ โ†’ dignity
  • ํŽธํ–ฅ โ†’ bias, skew
  • ์ฑ…์ž„ ์†Œ์žฌ โ†’ locus of responsibility
  • ๊ฐœ์ธ์ •๋ณด ๋ณดํ˜ธ โ†’ personal data protection
  • ํˆฌ๋ช…์„ฑ โ†’ transparency
  • ๊ทœ์ œ โ†’ regulation
  • ๊ณต์ •์„ฑ โ†’ fairness
  • ๊ฐ๋… โ†’ oversight
  • ์ž์œจ์„ฑ โ†’ autonomy
  • ์€ํํ•˜๋‹ค โ†’ to conceal
Tip ๐Ÿ’ก Most AI ethics vocabulary becomes powerful when you attach it to verbs like ๋ณดํ˜ธํ•˜๋‹ค (protect), ๋ณด์žฅํ•˜๋‹ค (guarantee), and ๋ช…ํ™•ํžˆ ํ•˜๋‹ค (clarify).

๐Ÿš€
Want more reps? Head into the Practice Hub below and train this vocabulary until your Korean sounds policy-panel ready. ๐Ÿ›๏ธ
ย 
๐Ÿ“ฆ
๐ŸŽฏ Practice Hub โ€” The Ethics of Artificial Intelligence
ย 
Welcome to your AI ethics practice zone, where robots behave politely and vocabulary carries legal consequences. ๐Ÿค– Youโ€™ll practice the core language of ์ธ๊ณต์ง€๋Šฅ ์œค๋ฆฌ (AI ethics), ์ฑ…์ž„ (responsibility), ํŽธํ–ฅ ๋ฐฉ์ง€ (bias prevention), and ์ธ๊ฐ„์„ฑ ๋ณดํ˜ธ (protection of humanity) through advanced, policy-style Korean sentences. โš–๏ธ
ย 
๐Ÿ”Š Audio reminder: Use
Korean Helper Widgets
Korean Helper Widgets
to hear any full sentence you want to shadow aloud.
ย 
๐Ÿ“ Instructions: Fill each blank with the Korean term or verb that best fits the sentence. Say the completed sentence aloud, check the Answer toggle, and notice whether the sentence is about fairness, responsibility, transparency, protection, or regulation.
ย 
Reference Hub โ€” Core nouns ๐Ÿง 
  • ์ธ๊ณต์ง€๋Šฅ ์œค๋ฆฌ โ†’ AI ethics
  • ์ฑ…์ž„ โ†’ responsibility
  • ์ฑ…์ž„ ์†Œ์žฌ โ†’ where responsibility lies
  • ํŽธํ–ฅ โ†’ bias
  • ํŽธํ–ฅ ๋ฐฉ์ง€ โ†’ bias prevention
  • ์ธ๊ฐ„์„ฑ ๋ณดํ˜ธ โ†’ protection of humanity
  • ์กด์—„์„ฑ โ†’ dignity
  • ์ž์œจ์„ฑ โ†’ autonomy
  • ํˆฌ๋ช…์„ฑ โ†’ transparency
  • ๊ณต์ •์„ฑ โ†’ fairness
ย 
Reference Hub โ€” Action verbs โš™๏ธ
  • ๋ณดํ˜ธํ•˜๋‹ค โ†’ to protect
  • ๋ณด์žฅํ•˜๋‹ค โ†’ to guarantee
  • ๋ฐฉ์ง€ํ•˜๋‹ค โ†’ to prevent
  • ํ™•๋ณดํ•˜๋‹ค โ†’ to secure
  • ๊ทœ์ œํ•˜๋‹ค โ†’ to regulate
  • ๊ฐ๋…ํ•˜๋‹ค โ†’ to oversee/supervise
  • ํ›ผ์†ํ•˜๋‹ค โ†’ to damage/undermine
  • ์ดˆ๋ž˜ํ•˜๋‹ค โ†’ to cause/bring about
  • ์€ํํ•˜๋‹ค โ†’ to conceal
  • ๋ช…ํ™•ํžˆ ํ•˜๋‹ค โ†’ to clarify
ย 

Exercise 1 โ€” Core Ethics Nouns โš–๏ธ

A. ์‚ฌํšŒ๋Š” ์ธ๊ณต์ง€๋Šฅ ๊ธฐ์ˆ ์˜ ๋ฐœ์ „๊ณผ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ์‚ฌ์ด์˜ ๊ท ํ˜•์„ ๊ณ ๋ฏผํ•ด์•ผ ํ•œ๋‹ค.
Society must consider the balance between AI development and ethics. ๐Ÿค–
Answer
์œค๋ฆฌ โ€” ๊ธฐ์ˆ  ๋ฐœ์ „๊ณผ ์œค๋ฆฌ ์‚ฌ์ด์˜ ๊ท ํ˜• is a natural policy-discussion frame.
B. ์ž๋™ ์ถ”์ฒœ ์‹œ์Šคํ…œ์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์€ ํŠน์ • ์ง‘๋‹จ์—๊ฒŒ ๋ถˆ๋ฆฌํ•œ ๊ฒฐ๊ณผ๋ฅผ ๋‚ณ์„ ์ˆ˜ ์žˆ๋‹ค.
The bias of an automated recommendation system may produce unfavorable results for a specific group. ๐Ÿ“ฑ
Answer
ํŽธํ–ฅ โ€” Use ํŽธํ–ฅ for skew or bias in systems and data.
C. ์ธ๊ณต์ง€๋Šฅ์˜ ์˜ค๋ฅ˜๋กœ ํ”ผํ•ด๊ฐ€ ๋ฐœ์ƒํ–ˆ์„ ๋•Œ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ์†Œ์žฌ๊ฐ€ ๋ถˆ๋ถ„๋ช…ํ•˜๋ฉด ๋ฌธ์ œ๊ฐ€ ์ปค์ง„๋‹ค.
When harm occurs because of an AI error, the problem grows if responsibility is unclear. ๐Ÿงพ
Answer
์ฑ…์ž„ โ€” Together, ์ฑ…์ž„ ์†Œ์žฌ means where responsibility lies.
D. ์ธ๊ฐ„์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ์ง€ํ‚ค๋Š” ๊ฒƒ์€ ๊ธฐ์ˆ  ๊ฐœ๋ฐœ์˜ ๊ธฐ๋ณธ ์ „์ œ์ด๋‹ค.
Protecting human dignity is a basic premise of technological development. ๐ŸŒฑ
Answer
์กด์—„์„ฑ โ€” ์ธ๊ฐ„์˜ ์กด์—„์„ฑ is the standard formal phrase.
E. AI๊ฐ€ ๋ชจ๋“  ๊ฒฐ์ •์„ ๋Œ€์‹ ํ•˜๋ฉด ์ธ๊ฐ„์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์ด ์•ฝํ•ด์งˆ ์ˆ˜ ์žˆ๋‹ค.
If AI makes every decision instead, human autonomy may weaken. ๐Ÿงญ
Answer
์ž์œจ์„ฑ โ€” ์ž์œจ์„ฑ is autonomy or self-directed choice.
F. ์˜์‚ฌ๊ฒฐ์ • ๊ณผ์ •์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์ด ๋‚ฎ์œผ๋ฉด ์‹œ๋ฏผ๋“ค์€ ๊ฒฐ๊ณผ๋ฅผ ์‹ ๋ขฐํ•˜๊ธฐ ์–ด๋ ต๋‹ค.
If transparency in the decision-making process is low, citizens find it difficult to trust the result. ๐Ÿ”
Answer
ํˆฌ๋ช…์„ฑ โ€” ํˆฌ๋ช…์„ฑ is transparency.
G. AI ์ฑ„์  ์‹œ์Šคํ…œ์€ ํ•™์ƒ๋“ค์—๊ฒŒ ๊ฐ™์€ ๊ธฐ์ค€์„ ์ ์šฉํ•˜๋Š” ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ํ™•๋ณดํ•ด์•ผ ํ•œ๋‹ค.
An AI grading system must secure fairness by applying the same standards to students. ๐Ÿ“
Answer
๊ณต์ •์„ฑ โ€” ๊ณต์ •์„ฑ means fairness.
H. ๋ฏผ๊ฐํ•œ ์ž๋ฃŒ๋ฅผ ์“ฐ๋Š” ์„œ๋น„์Šค์—์„œ๋Š” ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ๋ณดํ˜ธ๊ฐ€ ํ•„์ˆ˜์ ์ด๋‹ค.
In services that use sensitive materials, personal information protection is essential. ๐Ÿ”
Answer
๊ฐœ์ธ์ •๋ณด โ€” ๊ฐœ์ธ์ •๋ณด ๋ณดํ˜ธ means personal data/privacy protection.
I. ๊ณ ์œ„ํ—˜ ์ธ๊ณต์ง€๋Šฅ์—๋Š” ๋ฒ•์  ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์™€ ์ง€์†์ ์ธ ๊ฐ๋…์ด ํ•„์š”ํ•˜๋‹ค.
High-risk AI needs legal regulation and continuous oversight. ๐Ÿ›๏ธ
Answer
๊ทœ์ œ โ€” ๋ฒ•์  ๊ทœ์ œ means legal regulation.
J. AI์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์€ ํŽธ๋ฆฌํ•จ๋ณด๋‹ค ์‚ฌ๋žŒ์„ ์šฐ์„ ํ•ด์•ผ ํ•œ๋‹ค๋Š” ๊ด€์ ์„ ๋‹ด๊ณ  ์žˆ๋‹ค.
The protection of humanity in AI contains the view that people should come before convenience. ๐Ÿซถ
Answer
์ธ๊ฐ„์„ฑ ๋ณดํ˜ธ โ€” This phrase focuses on preserving human-centered values.
๐Ÿ“š
Vocabulary Box
  • ์œค๋ฆฌ โ†’ ethics
  • ํŽธํ–ฅ โ†’ bias
  • ์ฑ…์ž„ โ†’ responsibility
  • ์กด์—„์„ฑ โ†’ dignity
  • ์ž์œจ์„ฑ โ†’ autonomy
  • ํˆฌ๋ช…์„ฑ โ†’ transparency
  • ๊ณต์ •์„ฑ โ†’ fairness
  • ๊ฐœ์ธ์ •๋ณด โ†’ personal information
  • ๊ทœ์ œ โ†’ regulation
  • ์ธ๊ฐ„์„ฑ ๋ณดํ˜ธ โ†’ protection of humanity
  • ์˜์‚ฌ๊ฒฐ์ • ๊ณผ์ • โ†’ decision-making process
  • ๊ณ ์œ„ํ—˜ ์ธ๊ณต์ง€๋Šฅ โ†’ high-risk AI
Tip ๐Ÿ’ก Abstract nouns like ํˆฌ๋ช…์„ฑ, ๊ณต์ •์„ฑ, and ์ž์œจ์„ฑ often pair with ํ™•๋ณดํ•˜๋‹ค (secure) or ๋ณด์žฅํ•˜๋‹ค (guarantee).
ย 

Exercise 2 โ€” Responsibility and Accountability ๐Ÿงพ

A. ๊ฐœ๋ฐœ์ž๋Š” AI ๋ชจ๋ธ์˜ ํ•œ๊ณ„๋ฅผ ์‚ฌ์šฉ์ž์—๊ฒŒ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
Developers must disclose the limitations of an AI model to users. ๐Ÿ“ฃ
Answer
๊ณต๊ฐœํ•˜๋‹ค โ€” ๊ณต๊ฐœํ•˜๋‹ค means to disclose or make public.
B. ์šด์˜์ž๋Š” ์„œ๋น„์Šค ์žฅ์• ๋กœ ์ธํ•œ ํ”ผํ•ด์— ๋Œ€ํ•ด ์ฑ…์ž„์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ํ•œ๋‹ค.
Operators must take responsibility for damage caused by service failure. ๐Ÿ› ๏ธ
Answer
์ ธ์•ผ โ€” ์ฑ…์ž„์„ ์ง€๋‹ค means to take responsibility; ์ ธ์•ผ ํ•œ๋‹ค means must take.
C. ์ž๋™ํ™”๋œ ์‹ฌ์‚ฌ์—์„œ ์˜ค๋ฅ˜๊ฐ€ ๋‚˜๋ฉด ์ฑ…์ž„ ์†Œ์žฌ๋ฅผ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ํ•ด์•ผ ํ•œ๋‹ค.
If an error occurs in automated screening, the locus of responsibility must be clarified. ๐Ÿ“‹
Answer
๋ช…ํ™•ํžˆ โ€” ๋ช…ํ™•ํžˆ ํ•˜๋‹ค means to clarify.
D. ๊ธฐ์—…์ด ์œ„ํ—˜์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•˜๋ฉด ์†Œ๋น„์ž์˜ ์‹ ๋ขฐ๋Š” ๋น ๋ฅด๊ฒŒ ๋ฌด๋„ˆ์ง„๋‹ค.
If a company conceals risk, consumer trust collapses quickly. ๐Ÿ•ณ๏ธ
Answer
์€ํํ•˜๋‹ค โ€” ์€ํํ•˜๋‹ค means to conceal, often with a negative tone.
E. ๊ณต๊ณต๊ธฐ๊ด€์€ AI ๋„์ž… ์ „์— ์˜ํ–ฅ ํ‰๊ฐ€๋ฅผ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
Public institutions must conduct an impact assessment before introducing AI. ๐Ÿข
Answer
์‹ค์‹œํ•˜๋‹ค โ€” ํ‰๊ฐ€๋ฅผ ์‹ค์‹œํ•˜๋‹ค means to conduct an evaluation.
F. ํ”ผํ•ด์ž๊ฐ€ ๊ตฌ์ œ๋ฅผ ๋ฐ›์„ ์ˆ˜ ์žˆ๋„๋ก ์ ˆ์ฐจ๋ฅผ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
Procedures must be prepared so victims can receive remedy. ๐Ÿง‘โ€โš–๏ธ
Answer
๋งˆ๋ จํ•˜๋‹ค โ€” ์ ˆ์ฐจ๋ฅผ ๋งˆ๋ จํ•˜๋‹ค means to prepare or establish procedures.
G. AI ์ •์ฑ…์€ ๊ฐœ๋ฐœ ์†๋„๋ณด๋‹ค ์‚ฌํšŒ์  ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ์šฐ์„ ํ•ด์•ผ ํ•œ๋‹ค.
AI policy should prioritize social responsibility over speed of development. ๐Ÿ›๏ธ
Answer
์ฑ…์ž„ โ€” ์‚ฌํšŒ์  ์ฑ…์ž„ means social responsibility.
H. ์˜ค๋ฅ˜๊ฐ€ ๋ฐ˜๋ณต๋˜๋ฉด ๊ฐ๋… ๊ธฐ๊ด€์ด ์กฐ์‚ฌ์— ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ์ˆ˜ ์žˆ๋‹ค.
If errors repeat, a supervisory agency may launch an investigation. ๐Ÿ”Ž
Answer
์ฐฉ์ˆ˜ํ•  โ€” ์กฐ์‚ฌ์— ์ฐฉ์ˆ˜ํ•˜๋‹ค means to begin/launch an investigation.
I. ์ฑ…๋ฌด๋ฅผ ๋‹คํ•˜์ง€ ์•Š์€ ๊ธฐ์—…์€ ๋ฒ•์  ์ œ์žฌ๋ฅผ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ์ˆ˜ ์žˆ๋‹ค.
A company that fails to fulfill its duty may face legal sanctions. โš–๏ธ
Answer
๋ฐ›์„ โ€” ์ œ์žฌ๋ฅผ ๋ฐ›๋‹ค means to receive/be subject to sanctions.
J. AI ์œค๋ฆฌ๋Š” ์ฑ…์ž„์„ ๊ธฐ์ˆ  ๋’ค์— ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ์•Š๋„๋ก ๋งŒ๋“ ๋‹ค.
AI ethics prevents responsibility from being hidden behind technology. ๐ŸŽญ
Answer
์ˆจ๊ธฐ์ง€ โ€” ์ˆจ๊ธฐ๋‹ค means to hide; this sentence criticizes responsibility-dodging.
๐Ÿ“š
Vocabulary Box
  • ๊ฐœ๋ฐœ์ž โ†’ developer
  • ์šด์˜์ž โ†’ operator
  • ํ•œ๊ณ„ โ†’ limitation
  • ๊ณต๊ฐœํ•˜๋‹ค โ†’ to disclose
  • ์ฑ…์ž„์„ ์ง€๋‹ค โ†’ to take responsibility
  • ์ฑ…์ž„ ์†Œ์žฌ โ†’ locus of responsibility
  • ๋ช…ํ™•ํžˆ ํ•˜๋‹ค โ†’ to clarify
  • ์€ํํ•˜๋‹ค โ†’ to conceal
  • ์˜ํ–ฅ ํ‰๊ฐ€ โ†’ impact assessment
  • ์‹ค์‹œํ•˜๋‹ค โ†’ to conduct
  • ์ ˆ์ฐจ๋ฅผ ๋งˆ๋ จํ•˜๋‹ค โ†’ to establish procedures
  • ์กฐ์‚ฌ์— ์ฐฉ์ˆ˜ํ•˜๋‹ค โ†’ to launch an investigation
  • ์ œ์žฌ๋ฅผ ๋ฐ›๋‹ค โ†’ to face sanctions
  • ์ˆจ๊ธฐ๋‹ค โ†’ to hide
Tip ๐Ÿ’ก Responsibility vocabulary often uses set phrases: ์ฑ…์ž„์„ ์ง€๋‹ค, ์ฑ…์ž„ ์†Œ์žฌ๋ฅผ ๋ฐํžˆ๋‹ค, and ์ฑ…๋ฌด๋ฅผ ๋‹คํ•˜๋‹ค.
ย 

Exercise 3 โ€” Bias Prevention and Fairness ๐Ÿ›ก๏ธ

A. ํ•™์Šต ๋ฐ์ดํ„ฐ๊ฐ€ ํ•œ์ชฝ์œผ๋กœ ์น˜์šฐ์น˜๋ฉด ์•Œ๊ณ ๋ฆฌ์ฆ˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์ด ๋ฐœ์ƒํ•  ์ˆ˜ ์žˆ๋‹ค.
If training data leans to one side, algorithmic bias may occur. ๐Ÿงฎ
Answer
ํŽธํ–ฅ โ€” ์•Œ๊ณ ๋ฆฌ์ฆ˜ ํŽธํ–ฅ is algorithmic bias.
B. ํŽธํ–ฅ ๋ฐฉ์ง€๋ฅผ ์œ„ํ•ด ๋ฐ์ดํ„ฐ์˜ ๋Œ€ํ‘œ์„ฑ์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
To prevent bias, the representativeness of data must be secured. ๐Ÿ“Š
Answer
ํ™•๋ณดํ•˜๋‹ค โ€” ๋Œ€ํ‘œ์„ฑ์„ ํ™•๋ณดํ•˜๋‹ค means to secure representativeness.
C. ๋ชจ๋ธ ํ‰๊ฐ€์—๋Š” ๋‹ค์–‘ํ•œ ์ง‘๋‹จ์˜ ๊ฒฐ๊ณผ๋ฅผ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•˜๋Š” ๊ณผ์ •์ด ํ•„์š”ํ•˜๋‹ค.
Model evaluation needs a process of comparing results from diverse groups. ๐Ÿ‘ฅ
Answer
๋น„๊ตํ•˜๋‹ค โ€” ๋น„๊ตํ•˜๋‹ค means to compare.
D. ๋ถˆ๊ณต์ •ํ•œ ๊ฒฐ๊ณผ๊ฐ€ ๋ฐœ๊ฒฌ๋˜๋ฉด ๊ธฐ์ค€์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
If unfair results are found, the standards must be adjusted. โš™๏ธ
Answer
์กฐ์ •ํ•˜๋‹ค โ€” ๊ธฐ์ค€์„ ์กฐ์ •ํ•˜๋‹ค means to adjust standards.
E. ์ฑ„์šฉ AI๋Š” ์ง€์›์ž์˜ ๋ฐฐ๊ฒฝ๋ณด๋‹ค ์ง๋ฌด ๋Šฅ๋ ฅ์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
Hiring AI should evaluate job ability rather than an applicantโ€™s background. ๐Ÿ’ผ
Answer
ํ‰๊ฐ€ํ•˜๋‹ค โ€” ๋Šฅ๋ ฅ์„ ํ‰๊ฐ€ํ•˜๋‹ค means to evaluate ability.
F. ์‚ฌํšŒ์  ์•ฝ์ž์—๊ฒŒ ๋ถˆ๋ฆฌํ•œ ๊ฒฐ๊ณผ๋ฅผ ๋‚ด๋Š” ์‹œ์Šคํ…œ์€ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ์žƒ๋Š”๋‹ค.
A system that produces unfavorable results for socially vulnerable groups loses fairness. ๐Ÿง‘โ€โš–๏ธ
Answer
๊ณต์ •์„ฑ โ€” This sentence centers on fairness.
G. ํŽธ๊ฒฌ์€ ์‚ฌ๋žŒ์˜ ํƒœ๋„์— ๊ฐ€๊น๊ณ , ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์€ ์ž๋ฃŒ๋‚˜ ํŒ๋‹จ์˜ ๊ธฐ์šธ์–ด์ง์— ๊ฐ€๊น๋‹ค.
Prejudice is closer to a personโ€™s attitude, while bias is closer to skew in data or judgment. ๐Ÿง 
Answer
ํŽธํ–ฅ โ€” This is the key distinction: ํŽธ๊ฒฌ = prejudice, ํŽธํ–ฅ = bias/skew.
H. ์˜๋ฃŒ AI๋Š” ํŠน์ • ์ง€์—ญ ํ™˜์ž๋ฅผ ๊ณผ์†Œ ๋Œ€ํ‘œํ•˜์ง€ ์•Š๋„๋ก ๋ฐ์ดํ„ฐ๋ฅผ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
Medical AI must supplement data so patients from certain regions are not underrepresented. ๐Ÿฅ
Answer
๋ณด์™„ํ•˜๋‹ค โ€” ๋ฐ์ดํ„ฐ๋ฅผ ๋ณด์™„ํ•˜๋‹ค means to supplement or improve data.
I. ์‹ฌ์‚ฌ ๊ธฐ์ค€์„ ๊ณต๊ฐœํ•˜๋ฉด ๊ฒฐ๊ณผ์— ๋Œ€ํ•œ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๊ฐ€ ๋†’์•„์ง„๋‹ค.
If screening standards are disclosed, trust in the results increases. ๐Ÿค
Answer
์‹ ๋ขฐ โ€” ์‹ ๋ขฐ๊ฐ€ ๋†’์•„์ง€๋‹ค means trust increases.
J. ํŽธํ–ฅ ๋ฐฉ์ง€๋Š” ๊ธฐ์ˆ  ๋ฌธ์ œ๊ฐ€ ์•„๋‹ˆ๋ผ ์‚ฌํšŒ์  ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์˜ ๋ฌธ์ œ์ด๊ธฐ๋„ ํ•˜๋‹ค.
Bias prevention is not only a technical issue but also an issue of social fairness. โš–๏ธ
Answer
๊ณต์ •์„ฑ โ€” This frames bias prevention as a fairness issue.
๐Ÿ“š
Vocabulary Box
  • ํ•™์Šต ๋ฐ์ดํ„ฐ โ†’ training data
  • ์น˜์šฐ์น˜๋‹ค โ†’ to lean/skew
  • ์•Œ๊ณ ๋ฆฌ์ฆ˜ ํŽธํ–ฅ โ†’ algorithmic bias
  • ๋Œ€ํ‘œ์„ฑ โ†’ representativeness
  • ํ™•๋ณดํ•˜๋‹ค โ†’ to secure
  • ๋น„๊ตํ•˜๋‹ค โ†’ to compare
  • ์กฐ์ •ํ•˜๋‹ค โ†’ to adjust
  • ํ‰๊ฐ€ํ•˜๋‹ค โ†’ to evaluate
  • ๊ณต์ •์„ฑ โ†’ fairness
  • ํŽธ๊ฒฌ โ†’ prejudice
  • ํŽธํ–ฅ โ†’ bias/skew
  • ๋ณด์™„ํ•˜๋‹ค โ†’ to supplement/improve
  • ์‹ ๋ขฐ โ†’ trust
Tip ๐Ÿ’ก Use ํŽธํ–ฅ for algorithmic or data bias; reserve ํŽธ๊ฒฌ for prejudice in human perception or attitude.
ย 

Exercise 4 โ€” Humanity, Privacy, and Oversight ๐Ÿซถ

A. ๊ฐ์ • ๋…ธ๋™์„ ๋Œ€์‹ ํ•˜๋Š” AI๋„ ์ธ๊ฐ„์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ํ›ผ์†ํ•ด์„œ๋Š” ์•ˆ ๋œ๋‹ค.
Even AI that replaces emotional labor must not damage human dignity. ๐Ÿซถ
Answer
์กด์—„์„ฑ โ€” The natural phrase is ์ธ๊ฐ„์˜ ์กด์—„์„ฑ.
B. ๋ฌด๋ถ„๋ณ„ํ•œ ๊ฐ์‹œ๋Š” ๊ฐœ์ธ์ •๋ณด ๋ณดํ˜ธ ์›์น™์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•  ์ˆ˜ ์žˆ๋‹ค.
Indiscriminate surveillance can damage the principle of personal information protection. ๐Ÿ”
Answer
ํ›ผ์†ํ•˜๋‹ค โ€” ์›์น™์„ ํ›ผ์†ํ•˜๋‹ค means to undermine/damage a principle.
C. AI๊ฐ€ ์กฐ์–ธ์„ ์ œ๊ณตํ•˜๋”๋ผ๋„ ์ตœ์ข… ํŒ๋‹จ์—๋Š” ์ธ๊ฐ„์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์ด ๋‚จ์•„ ์žˆ์–ด์•ผ ํ•œ๋‹ค.
Even if AI provides advice, human judgment must remain in the final decision. ๐Ÿง‘โ€โš–๏ธ
Answer
ํŒ๋‹จ โ€” This protects human judgment in decision-making.
D. ํ™˜์ž์˜ ์ •๋ณด๋Š” ์—ฐ๊ตฌ ๋ชฉ์ ์ด๋ผ๋„ ์—„๊ฒฉํ•˜๊ฒŒ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋˜์–ด์•ผ ํ•œ๋‹ค.
Patient information must be strictly protected even for research purposes. ๐Ÿฅ
Answer
๋ณดํ˜ธ๋˜๋‹ค โ€” Passive ๋ณดํ˜ธ๋˜๋‹ค means to be protected.
E. ์ธ๊ฐ„์„ฑ ๋ณดํ˜ธ๋Š” ํšจ์œจ์„ฑ๋ณด๋‹ค ์‚ฌ๋žŒ์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ๋จผ์ € ์ƒ๊ฐํ•œ๋‹ค.
The protection of humanity considers peopleโ€™s value before efficiency. ๐ŸŒฑ
Answer
๊ฐ€์น˜ โ€” ์‚ฌ๋žŒ์˜ ๊ฐ€์น˜ is a broad, human-centered phrase.
F. ์ž๋™ํ™”๊ฐ€ ์ปค์งˆ์ˆ˜๋ก ์ธ๊ฐ„์˜ ๊ฐ๋… ์ฒด๊ณ„๋ฅผ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
As automation grows, human oversight systems must be strengthened. ๐Ÿ‘€
Answer
๊ฐ•ํ™”ํ•˜๋‹ค โ€” ์ฒด๊ณ„๋ฅผ ๊ฐ•ํ™”ํ•˜๋‹ค means to strengthen a system.
G. ์„ค๋ช… ์—†์ด ์ ์ˆ˜๋งŒ ์ œ์‹œํ•˜๋Š” AI๋Š” ์‚ฌ์šฉ์ž์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ์ œํ•œํ•  ์ˆ˜ ์žˆ๋‹ค.
AI that presents only scores without explanation can limit the userโ€™s autonomy. ๐Ÿ“Š
Answer
์ž์œจ์„ฑ โ€” A lack of explanation can weaken meaningful choice.
H. ๋Œ๋ด„ ๋กœ๋ด‡์€ ํŽธ๋ฆฌํ•จ๋ฟ ์•„๋‹ˆ๋ผ ๊ด€๊ณ„์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋„ ๊ณ ๋ คํ•ด์•ผ ํ•œ๋‹ค.
Care robots must consider not only convenience but also the humanity of relationships. ๐Ÿค
Answer
์ธ๊ฐ„์„ฑ โ€” Here ๊ด€๊ณ„์˜ ์ธ๊ฐ„์„ฑ means the humane quality of relationships.
I. ๋ฐ์ดํ„ฐ ์ˆ˜์ง‘์€ ํ•„์š”ํ•œ ๋ฒ”์œ„ ์•ˆ์—์„œ๋งŒ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋˜์–ด์•ผ ํ•œ๋‹ค.
Data collection should be allowed only within the necessary scope. ๐Ÿ“
Answer
ํ—ˆ์šฉ๋˜๋‹ค โ€” ํ—ˆ์šฉ๋˜๋‹ค means to be allowed.
J. AI ์œค๋ฆฌ๋Š” ๊ธฐ์ˆ ์ด ์‚ฌ๋žŒ์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•˜์ง€ ์•Š๋„๋ก ๊ฒฝ๊ณ„ํ•œ๋‹ค.
AI ethics guards against technology alienating people. ๐ŸŒซ๏ธ
Answer
์†Œ์™ธ์‹œํ‚ค๋‹ค โ€” ์†Œ์™ธ์‹œํ‚ค๋‹ค means to alienate or marginalize.
๐Ÿ“š
Vocabulary Box
  • ๊ฐ์ • ๋…ธ๋™ โ†’ emotional labor
  • ์กด์—„์„ฑ โ†’ dignity
  • ๋ฌด๋ถ„๋ณ„ํ•œ ๊ฐ์‹œ โ†’ indiscriminate surveillance
  • ๊ฐœ์ธ์ •๋ณด ๋ณดํ˜ธ โ†’ personal information protection
  • ํ›ผ์†ํ•˜๋‹ค โ†’ to damage/undermine
  • ํŒ๋‹จ โ†’ judgment
  • ๋ณดํ˜ธ๋˜๋‹ค โ†’ to be protected
  • ๊ฐ€์น˜ โ†’ value
  • ๊ฐ๋… ์ฒด๊ณ„ โ†’ oversight system
  • ๊ฐ•ํ™”ํ•˜๋‹ค โ†’ to strengthen
  • ์ž์œจ์„ฑ โ†’ autonomy
  • ์ธ๊ฐ„์„ฑ โ†’ humanity
  • ํ—ˆ์šฉ๋˜๋‹ค โ†’ to be allowed
  • ์†Œ์™ธ์‹œํ‚ค๋‹ค โ†’ to alienate
Tip ๐Ÿ’ก ๋ณดํ˜ธ prevents harm; ๋ณด์žฅ guarantees a right or condition. Choose the verb based on the ethical relationship.
ย 

Exercise 5 โ€” Mixed Mastery + Pitfall Patrol ๐Ÿ›ก๏ธ

A. AI๊ฐ€ ์Šค์Šค๋กœ ๋ฒ•์  ์ฑ…์ž„์„ ์ง„๋‹ค๋Š” ํ‘œํ˜„๋ณด๋‹ค ๊ฐœ๋ฐœ์ž์™€ ์šด์˜์ž์˜ ์ฑ…์ž„์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ํ•œ๋‹ค๊ฐ€ ๋” ์ž์—ฐ์Šค๋Ÿฝ๋‹ค.
Rather than saying AI itself bears legal responsibility, it is more natural to say responsibility of developers and operators must be clarified. โš–๏ธ
Answer
๋ช…ํ™•ํžˆ โ€” โŒ assigning responsibility to AI itself โ†’ โœ… clarifying human/institutional responsibility.
B. ์•Œ๊ณ ๋ฆฌ์ฆ˜ ๋ฌธ์ œ์—๋Š” ํŽธ๊ฒฌ๋ณด๋‹ค ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์ด๋ผ๋Š” ๋ง์ด ๋” ์ •ํ™•ํ•œ ๊ฒฝ์šฐ๊ฐ€ ๋งŽ๋‹ค.
For algorithm problems, โ€œbiasโ€ is often more accurate than โ€œprejudice.โ€ ๐Ÿงฎ
Answer
ํŽธํ–ฅ โ€” โŒ ํŽธ๊ฒฌ ๋ฐฉ์ง€ for data fairness โ†’ โœ… ํŽธํ–ฅ ๋ฐฉ์ง€.
C. ์ธ๊ฐ„์„ฑ์€ ๋ณดํ†ต ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•˜๊ณ , ๊ถŒ๋ฆฌ๋‚˜ ์กด์—„์„ฑ์€ ๋ณด์žฅํ•œ๋‹ค๊ณ  ๋งํ•œ๋‹ค.
We usually say humanity is protected, while rights or dignity are guaranteed. ๐Ÿซถ
Answer
๋ณดํ˜ธํ•˜๋‹ค โ€” โŒ defaulting to ์ธ๊ฐ„์„ฑ ๋ณด์žฅ โ†’ โœ… ์ธ๊ฐ„์„ฑ ๋ณดํ˜ธ.
D. ์„ค๋ช… ๊ฐ€๋Šฅํ•œ ์ธ๊ณต์ง€๋Šฅ์€ ๋‹จ์ˆœํžˆ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ AI๋ณด๋‹ค ๋” ๊ตฌ์ฒด์ ์ธ ํ‘œํ˜„์ด๋‹ค.
Explainable AI is more specific than simply โ€œtransparent AI.โ€ ๐Ÿ”
Answer
ํˆฌ๋ช…ํ•œ โ€” โŒ only saying ํˆฌ๋ช…ํ•œ AI โ†’ โœ… using ์„ค๋ช… ๊ฐ€๋Šฅํ•œ ์ธ๊ณต์ง€๋Šฅ when explainability is meant.
E. ์ ์ ˆํ•œ ๊ทœ์ œ๋Š” ํ˜์‹ ์„ ๋ง‰๊ธฐ๋งŒ ํ•˜๋Š” ๊ฒƒ์ด ์•„๋‹ˆ๋ผ ์•ˆ์ „๊ณผ ํ˜์‹ ์˜ ๊ท ํ˜•์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•œ๋‹ค.
Appropriate regulation does not merely block innovation; it promotes balance between safety and innovation. ๐Ÿ›๏ธ
Answer
๋„๋ชจํ•˜๋‹ค โ€” โŒ regulation as only anti-innovation โ†’ โœ… regulation as balance-seeking.
F. ์œ„ํ—˜์„ ์ˆจ๊ธฐ๋Š” ํ–‰์œ„๋Š” ๊ณต๊ฐœ๊ฐ€ ์•„๋‹ˆ๋ผ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์— ๊ฐ€๊น๋‹ค.
Hiding risk is closer to concealment than disclosure. ๐Ÿšจ
Answer
์€ํ โ€” ์€ํ is concealment; it carries a critical tone.
G. AI๊ฐ€ ์‚ฌ๋žŒ์˜ ๊ฒฐ์ •์„ ๋Œ€์‹ ํ• ์ˆ˜๋ก ์ธ๊ฐ„์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ์ง€ํ‚ค๋Š” ์žฅ์น˜๊ฐ€ ์ค‘์š”ํ•ด์ง„๋‹ค.
As AI replaces human decisions, mechanisms that protect human autonomy become more important. ๐Ÿงญ
Answer
์ž์œจ์„ฑ โ€” The ethical issue is meaningful human choice, not just convenience.
H. ๋ถˆํˆฌ๋ช…ํ•œ ๋ชจ๋ธ์€ ๊ฒฐ๊ณผ์— ๋Œ€ํ•œ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋ฅผ ๋–จ์–ด๋œจ๋ฆด ์ˆ˜ ์žˆ๋‹ค.
An opaque model may lower trust in the result. ๐ŸŒซ๏ธ
Answer
์‹ ๋ขฐ โ€” Opaque decisions weaken trust.
I. ์‚ฌํšŒ์  ์•ฝ์ž๋ฅผ ๋ถˆ๋ฆฌํ•˜๊ฒŒ ๋งŒ๋“œ๋Š” ๊ฒฐ๊ณผ๋Š” AI์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ๋ฌธ์ œ๋กœ ์ด์–ด์ง„๋‹ค.
Results that disadvantage socially vulnerable groups lead to a fairness problem in AI. ๐Ÿง‘โ€โš–๏ธ
Answer
๊ณต์ •์„ฑ โ€” This directly concerns fairness.
J. ์ธ๊ณต์ง€๋Šฅ ์œค๋ฆฌ์˜ ํ•ต์‹ฌ์€ ๊ธฐ์ˆ ์„ ๋ฉˆ์ถ”๋Š” ๊ฒƒ์ด ์•„๋‹ˆ๋ผ ์‚ฌ๋žŒ์„ ์ค‘์‹ฌ์— ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ๊ฒƒ์ด๋‹ค.
The core of AI ethics is not stopping technology, but placing people at the center. ๐ŸŽฏ
Answer
๋‘๋Š” โ€” ์ค‘์‹ฌ์— ๋‘๋‹ค means to place at the center. Very handy phrase; tiny grammar, big ethical cape.
๐Ÿ“š
Vocabulary Box
  • ๋ช…ํ™•ํžˆ ํ•˜๋‹ค โ†’ to clarify
  • ํŽธ๊ฒฌ โ†’ prejudice
  • ํŽธํ–ฅ โ†’ bias/skew
  • ๋ณดํ˜ธํ•˜๋‹ค โ†’ to protect
  • ํˆฌ๋ช…ํ•œ โ†’ transparent
  • ์„ค๋ช… ๊ฐ€๋Šฅํ•œ ์ธ๊ณต์ง€๋Šฅ โ†’ explainable AI
  • ๋„๋ชจํ•˜๋‹ค โ†’ to promote/seek
  • ์€ํ โ†’ concealment
  • ์ž์œจ์„ฑ โ†’ autonomy
  • ์‹ ๋ขฐ โ†’ trust
  • ๊ณต์ •์„ฑ โ†’ fairness
  • ์ค‘์‹ฌ์— ๋‘๋‹ค โ†’ to place at the center
  • ์‚ฌํšŒ์  ์•ฝ์ž โ†’ socially vulnerable group
Tip ๐Ÿ’ก For pitfall patrol, ask: Is this about a human attitude, a data skew, an institutional duty, or a right that must be guaranteed?
ย 
๐ŸŽ‰ Excellent work. You can now discuss AI ethics in Korean with words that point to real-world accountability: ์ฑ…์ž„ ์†Œ์žฌ, ํŽธํ–ฅ ๋ฐฉ์ง€, ํˆฌ๋ช…์„ฑ, ๊ณต์ •์„ฑ, and ์ธ๊ฐ„์„ฑ ๋ณดํ˜ธ. The robots may still be mysterious, but your vocabulary is no longer hiding in the server room. ๐Ÿค–๐Ÿ”