๐Ÿ›ฐ๏ธ

The Reliability of Automated Systems

Agent Progress
Built
Keywords
์ž๋™ํ™” ์‹œ์Šคํ…œ์˜ ์‹ ๋ขฐ์„ฑ / ์•Œ๊ณ ๋ฆฌ์ฆ˜ / ์˜ค๋ฅ˜ ๊ฐ€๋Šฅ์„ฑ / ์•ˆ์ „์„ฑ
Lesson #
S
Done
Section
Advanced Vocabulary
๐Ÿ›ฐ๏ธ
TL;DR ๐Ÿฆ„
  • ๐Ÿงญ ์‹ ๋ขฐ์„ฑ means reliability: a system keeps working consistently, safely, and predictably.
  • โš™๏ธ Automated-system Korean often combines ์•Œ๊ณ ๋ฆฌ์ฆ˜, ์˜ค๋ฅ˜ ๊ฐ€๋Šฅ์„ฑ, ์•ˆ์ „์„ฑ, ๊ฒ€์ฆ, ํˆฌ๋ช…์„ฑ, and ์ฑ…์ž„ ์†Œ์žฌ.
  • โš–๏ธ Strong advanced wording balances convenience with risk: not โ€œautomation is perfect,โ€ but โ€œreliability must be secured and verified.โ€
๐ŸŽง
๐ŸŽง Iru Helper Note โ€” Open
Korean Helper Widgets
Korean Helper Widgets
whenever you want to hear a phrase like ์ž๋™ํ™” ์‹œ์Šคํ…œ์˜ ์‹ ๋ขฐ์„ฑ or ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ํŒ๋‹จ ๊ธฐ์ค€ read aloud. Paste the full Korean sentence into the widget and listen for the rhythm of long noun phrases.
ย 

๐ŸŒˆ The Big Picture

Automated systems are everywhere: navigation apps ๐Ÿš—, hospital screening tools ๐Ÿฅ, payment systems ๐Ÿ’ณ, recommendation feeds ๐Ÿ“ฑ, factory robots ๐Ÿญ, and customer-service chat systems ๐Ÿ’ฌ. Advanced Korean discusses these systems with abstract nouns, so the lesson goal is to make those noun-heavy chunks feel natural instead of foggy.
The central word is ์‹ ๋ขฐ์„ฑ = reliability. It is not the same as personal trust. ์‹ ๋ขฐ can mean emotional or social trust; ์‹ ๋ขฐ์„ฑ means the quality of being dependable, especially for data, methods, systems, evidence, and technology.
For automated systems, reliability usually depends on four connected ideas: accurate decisions, low error risk, verified safety, and clear responsibility when something goes wrong. In Korean, those ideas often appear as compact noun phrases like ์˜ค๋ฅ˜ ๋ฐœ์ƒ ๊ฐ€๋Šฅ์„ฑ (possibility of error occurring) or ์ฑ…์ž„ ์†Œ์žฌ (where responsibility lies).

๐ŸŽจ The Core Reliability Vocabulary Family

  • ์ž๋™ํ™” ์‹œ์Šคํ…œ = automated system. Use it with ๋„์ž…ํ•˜๋‹ค (to introduce/adopt), ์šด์˜ํ•˜๋‹ค (to operate), and ๊ด€๋ฆฌํ•˜๋‹ค (to manage).
  • ์‹ ๋ขฐ์„ฑ = reliability. Natural partners: ์‹ ๋ขฐ์„ฑ์„ ๋†’์ด๋‹ค (improve reliability), ์‹ ๋ขฐ์„ฑ์„ ํ™•๋ณดํ•˜๋‹ค (secure reliability), ์‹ ๋ขฐ์„ฑ์„ ๊ฒ€์ฆํ•˜๋‹ค (verify reliability).
  • ์•Œ๊ณ ๋ฆฌ์ฆ˜ = algorithm. Pair it with ํŒ๋‹จ ๊ธฐ์ค€ (decision criteria), ํŽธํ–ฅ (bias), ์™œ๊ณก (distortion), and ์ž‘๋™ ๋ฐฉ์‹ (operating method).
  • ์˜ค๋ฅ˜ ๊ฐ€๋Šฅ์„ฑ = possibility of error. For a slightly more formal phrase, use ์˜ค๋ฅ˜ ๋ฐœ์ƒ ๊ฐ€๋Šฅ์„ฑ = possibility that an error will occur.
  • ์•ˆ์ „์„ฑ = safety as a quality/degree. Use it for technical systems: ์•ˆ์ „์„ฑ์„ ํ™•๋ณดํ•˜๋‹ค (secure safety), ์•ˆ์ „์„ฑ์„ ๊ฒ€์ฆํ•˜๋‹ค (verify safety).
  • ํˆฌ๋ช…์„ฑ = transparency. This often means users can understand why a system produced a result.
  • ์ฑ…์ž„ ์†Œ์žฌ = responsibility/accountability: who is responsible when harm or error occurs.

๐Ÿ› ๏ธ How They Work in a Sentence

  • ์ž๋™ํ™” ์‹œ์Šคํ…œ์˜ ์‹ ๋ขฐ์„ฑ์„ ํ™•๋ณดํ•˜๋ ค๋ฉด ์˜ค๋ฅ˜ ๊ฐ€๋Šฅ์„ฑ์„ ์‚ฌ์ „์— ์ค„์—ฌ์•ผ ํ•œ๋‹ค. โ€” To secure the reliability of an automated system, the possibility of errors must be reduced in advance. ๐Ÿงฏ
    • ํ™•๋ณดํ•˜๋‹ค means โ€œsecure/ensure,โ€ and ์‚ฌ์ „์— means โ€œin advance.โ€
  • ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ํŒ๋‹จ ๊ธฐ์ค€์ด ๋ถˆํˆฌ๋ช…ํ•˜๋ฉด ์‚ฌ์šฉ์ž๋Š” ๊ฒฐ๊ณผ๋ฅผ ์‹ ๋ขฐํ•˜๊ธฐ ์–ด๋ ต๋‹ค. โ€” If an algorithmโ€™s decision criteria are opaque, users have difficulty trusting the result. ๐Ÿ”
    • ํŒ๋‹จ ๊ธฐ์ค€ = decision criteria; ๋ถˆํˆฌ๋ช…ํ•˜๋‹ค = opaque/not transparent.
  • ์•ˆ์ „์„ฑ์ด ์ถฉ๋ถ„ํžˆ ๊ฒ€์ฆ๋˜์ง€ ์•Š์€ ๊ธฐ์ˆ ์„ ์„ฑ๊ธ‰ํ•˜๊ฒŒ ๋„์ž…ํ•˜๋ฉด ์ฑ…์ž„ ์†Œ์žฌ๊ฐ€ ๋ถˆ๋ถ„๋ช…ํ•ด์งˆ ์ˆ˜ ์žˆ๋‹ค. โ€” If a technology whose safety has not been sufficiently verified is introduced hastily, responsibility may become unclear. โš ๏ธ
    • ์„ฑ๊ธ‰ํ•˜๊ฒŒ = hastily; ๋ถˆ๋ถ„๋ช…ํ•ด์ง€๋‹ค = become unclear.

๐Ÿงฉ General Formation Rule

Advanced technical Korean often builds meaning by stacking nouns:
  • ์ž๋™ํ™” ์‹œ์Šคํ…œ + ์˜ + ์‹ ๋ขฐ์„ฑ โ†’ ์ž๋™ํ™” ์‹œ์Šคํ…œ์˜ ์‹ ๋ขฐ์„ฑ (the reliability of automated systems)
  • ์•Œ๊ณ ๋ฆฌ์ฆ˜ + ์˜ + ํŒ๋‹จ ๊ธฐ์ค€ โ†’ ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ํŒ๋‹จ ๊ธฐ์ค€ (an algorithmโ€™s decision criteria)
  • ์˜ค๋ฅ˜ + ๋ฐœ์ƒ + ๊ฐ€๋Šฅ์„ฑ โ†’ ์˜ค๋ฅ˜ ๋ฐœ์ƒ ๊ฐ€๋Šฅ์„ฑ (possibility of error occurrence)
  • ์ฑ…์ž„ + ์†Œ์žฌ โ†’ ์ฑ…์ž„ ์†Œ์žฌ (where responsibility lies)
The grammar is ordinary noun linking, not an exception. The advanced skill is choosing natural pairings: ์‹ ๋ขฐ์„ฑ์„ ํ™•๋ณดํ•˜๋‹ค, ์•ˆ์ „์„ฑ์„ ๊ฒ€์ฆํ•˜๋‹ค, ์˜ค๋ฅ˜๋ฅผ ๋ฐฉ์ง€ํ•˜๋‹ค, ํŒ๋‹จ ๊ธฐ์ค€์„ ๊ณต๊ฐœํ•˜๋‹ค.

๐Ÿ“š Cross-check note

The key Korean vocabulary, noun pairings, and technical-style collocations in this lesson were cross-checked against usage patterns available through the National Institute of Korean Language and ๊ตญ๋ฆฝ๊ตญ์–ด์› resources at korean.go.kr. โœ…

๐Ÿชž Synonym Alert

  • ์‹ ๋ขฐ = trust, confidence; often personal or social. ์‹ ๋ขฐ์„ฑ = reliability as an objective quality.
  • ์•ˆ์ „ = safety in a broad sense. ์•ˆ์ „์„ฑ = degree/quality of safety, especially in technical or policy writing.
  • ์˜ค๋ฅ˜ = error in data, logic, calculation, or a system. ์‹ค์ˆ˜ = a human mistake.
  • ํŽธํ–ฅ = bias. ์™œ๊ณก = distortion. Bias can cause distorted results.
  • ์ฑ…์ž„ = responsibility. ์ฑ…์ž„ ์†Œ์žฌ = where responsibility lies, especially after a problem.

โš ๏ธ Pitfalls to Dodge

  • โŒ ์ž๋™ํ™” ์‹œ์Šคํ…œ์„ ๋ฏฟ์œผ๋ฉด ๋œ๋‹ค. โ†’ โœ… ์ž๋™ํ™” ์‹œ์Šคํ…œ์˜ ์‹ ๋ขฐ์„ฑ์„ ๊ฒ€์ฆํ•ด์•ผ ํ•œ๋‹ค. (Advanced writing needs verification, not blind trust.)
  • โŒ ์‚ฌ๋žŒ์ด ์•ˆ ํ•˜๋‹ˆ๊นŒ ์˜ค๋ฅ˜๊ฐ€ ์—†๋‹ค. โ†’ โœ… ์ž๋™ํ™”๋˜์–ด๋„ ์˜ค๋ฅ˜ ๊ฐ€๋Šฅ์„ฑ์€ ๋‚จ์•„ ์žˆ๋‹ค. (Automation reduces some errors but does not remove all risk.)
  • โŒ ์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ํ•ญ์ƒ ๊ฐ๊ด€์ ์ด๋‹ค. โ†’ โœ… ๋ฐ์ดํ„ฐ๊ฐ€ ํŽธํ–ฅ๋˜๋ฉด ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ํŒ๋‹จ๋„ ์™œ๊ณก๋  ์ˆ˜ ์žˆ๋‹ค. (Algorithms can reflect biased data.)
  • โŒ ์•ˆ์ „๋งŒ ํ™•์ธํ•˜๋ฉด ๋œ๋‹ค. โ†’ โœ… ์•ˆ์ „์„ฑ๊ณผ ์‹ ๋ขฐ์„ฑ์„ ํ•จ๊ป˜ ๊ฒ€์ฆํ•ด์•ผ ํ•œ๋‹ค. (Technical evaluation needs both safety and reliability.)
  • โŒ ๋ฌธ์ œ๊ฐ€ ์ƒ๊ธฐ๋ฉด ๋ˆ„๊ตฐ๊ฐ€ ์ฑ…์ž„์ง„๋‹ค. โ†’ โœ… ๋ฌธ์ œ๊ฐ€ ๋ฐœ์ƒํ–ˆ์„ ๋•Œ ์ฑ…์ž„ ์†Œ์žฌ๋ฅผ ๋ช…ํ™•ํžˆ ํ•ด์•ผ ํ•œ๋‹ค. (Accountability must be defined clearly.)
Nice work โ€” this vocabulary lets you discuss technology without sounding like a robot describing another robot. ๐Ÿค–โœจ
ย 

Try this exercise

Practice completing advanced Korean sentences about automated-system reliability. Choose the most natural word or phrase for each blank, say the full sentence aloud, then open the Answer toggle to check the nuance.
1. ์ž๋™ํ™” ์‹œ์Šคํ…œ์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ํ™•๋ณดํ•˜๋ ค๋ฉด ๋ฐ˜๋ณต์ ์ธ ๊ฒ€์ฆ์ด ํ•„์š”ํ•˜๋‹ค.
The reliability of automated systems requires repeated verification. ๐Ÿงญ
Answer
์‹ ๋ขฐ์„ฑ โ€” ์‹ ๋ขฐ์„ฑ์„ ํ™•๋ณดํ•˜๋‹ค is the natural technical pairing for โ€œsecure reliability.โ€
2. ๋ฐ์ดํ„ฐ๊ฐ€ ํ•œ์ชฝ์œผ๋กœ ์น˜์šฐ์น˜๋ฉด ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์ด ๋ฐœ์ƒํ•  ์ˆ˜ ์žˆ๋‹ค.
If data leans to one side, algorithmic bias can occur. โš–๏ธ
Answer
ํŽธํ–ฅ โ€” ํŽธํ–ฅ means bias; biased data can influence an algorithm.
3. ์˜๋ฃŒ ๋ถ„์•ผ์—์„œ๋Š” ๊ธฐ์ˆ ์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ๋จผ์ € ๊ฒ€์ฆํ•ด์•ผ ํ•œ๋‹ค.
In medicine, the safety of a technology must be verified first. ๐Ÿฅ
Answer
์•ˆ์ „์„ฑ โ€” ์•ˆ์ „์„ฑ is the quality/degree of safety in technical contexts.
4. ํŒ๋‹จ ๊ณผ์ •์ด ๋ถˆํˆฌ๋ช…ํ•˜๋ฉด ๊ฒฐ๊ณผ์— ๋Œ€ํ•œ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๊ฐ€ ๋‚ฎ์•„์ง„๋‹ค.
If the decision process is opaque, trust in the result decreases. ๐Ÿ”
Answer
์‹ ๋ขฐ โ€” Here the sentence describes user trust, so ์‹ ๋ขฐ is natural.
5. ์˜ค๋ฅ˜๋ฅผ ์ค„์ด๊ธฐ ์œ„ํ•ด ์‚ฌ์ „ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ์ ˆ์ฐจ๋ฅผ ๊ฐ•ํ™”ํ–ˆ๋‹ค.
They strengthened pre-check procedures to reduce errors. ๐Ÿงช
Answer
์ ๊ฒ€ โ€” ์‚ฌ์ „ ์ ๊ฒ€ means pre-checking/inspection.
6. ๋ฌธ์ œ๊ฐ€ ๋ฐœ์ƒํ•˜๋ฉด ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋ฅผ ๋ช…ํ™•ํžˆ ํ•ด์•ผ ํ•œ๋‹ค.
If a problem occurs, accountability must be clarified. ๐Ÿ“Œ
Answer
์ฑ…์ž„ ์†Œ์žฌ โ€” This phrase asks where responsibility lies.
7. ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ํŒ๋‹จ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ๊ณต๊ฐœํ•ด์•ผ ์‚ฌ์šฉ์ž๊ฐ€ ๊ฒฐ๊ณผ๋ฅผ ์ดํ•ดํ•  ์ˆ˜ ์žˆ๋‹ค.
The algorithmโ€™s decision criteria must be disclosed so users can understand the result. ๐Ÿง 
Answer
๊ธฐ์ค€ โ€” ํŒ๋‹จ ๊ธฐ์ค€ = decision criteria.
8. ๊ฒ€์ฆ๋˜์ง€ ์•Š์€ ๊ธฐ์ˆ ์„ ์„ฑ๊ธ‰ํ•˜๊ฒŒ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•˜๋ฉด ์œ„ํ—˜์ด ์ปค์ง„๋‹ค.
If an unverified technology is adopted hastily, risk increases. ๐Ÿšง
Answer
๋„์ž… โ€” ๊ธฐ์ˆ ์„ ๋„์ž…ํ•˜๋‹ค means to introduce/adopt technology.
9. ์ž๋™ํ™”๋˜์–ด๋„ ์˜ค๋ฅ˜ ๊ฐ€๋Šฅ์„ฑ์„ ์™„์ „ํžˆ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•  ์ˆ˜๋Š” ์—†๋‹ค.
Even if something is automated, the possibility of error cannot be completely eliminated. ๐Ÿงฏ
Answer
์ œ๊ฑฐ โ€” ๊ฐ€๋Šฅ์„ฑ์„ ์ œ๊ฑฐํ•˜๋‹ค means to eliminate a possibility.
10. ์‹œ์Šคํ…œ์˜ ์ž‘๋™ ๋ฐฉ์‹์— ๋Œ€ํ•œ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์ด ๋†’์„์ˆ˜๋ก ์‚ฌ์šฉ์ž๋Š” ๊ฒฐ๊ณผ๋ฅผ ๋‚ฉ๋“ํ•˜๊ธฐ ์‰ฝ๋‹ค.
The higher the transparency of the systemโ€™s operating method, the easier it is for users to accept the result. ๐ŸชŸ
Answer
ํˆฌ๋ช…์„ฑ โ€” ํˆฌ๋ช…์„ฑ means transparency, especially in decision processes.
๐Ÿ“š
Vocabulary Box
  • ์‹ ๋ขฐ์„ฑ โ†’ reliability
  • ํŽธํ–ฅ โ†’ bias
  • ์•ˆ์ „์„ฑ โ†’ safety as a quality
  • ์‹ ๋ขฐ โ†’ trust
  • ์ ๊ฒ€ โ†’ check/inspection
  • ์ฑ…์ž„ ์†Œ์žฌ โ†’ where responsibility lies
  • ํŒ๋‹จ ๊ธฐ์ค€ โ†’ decision criteria
  • ๋„์ž… โ†’ introduction/adoption
  • ์ œ๊ฑฐ โ†’ elimination
  • ํˆฌ๋ช…์„ฑ โ†’ transparency
Tip ๐Ÿ’ก Match abstract nouns with their natural verbs: ํ™•๋ณดํ•˜๋‹ค, ๊ฒ€์ฆํ•˜๋‹ค, ๊ณต๊ฐœํ•˜๋‹ค, ๋ช…ํ™•ํžˆ ํ•˜๋‹ค.

๐Ÿš€
Want more reps? Visit the Practice Hub when you're ready!
ย 
๐Ÿ“ฆ
๐ŸŽฏ Practice Hub โ€” The Reliability of Automated Systems
ย 
Welcome to your reliability workout for automated systems. ๐Ÿ›ฐ๏ธ Youโ€™ll practice the vocabulary that makes advanced Korean sound precise: ์‹ ๋ขฐ์„ฑ, ์•ˆ์ „์„ฑ, ์˜ค๋ฅ˜ ๊ฐ€๋Šฅ์„ฑ, ํˆฌ๋ช…์„ฑ, and ์ฑ…์ž„ ์†Œ์žฌ. By the end, your sentences will sound less like โ€œtechnology good/badโ€ and more like โ€œrisk, verification, and accountability are connected.โ€ Very elegant. Very serious. Still a little sparkly. โœจ
ย 
๐Ÿ”Š Audio reminder: Use
Korean Helper Widgets
Korean Helper Widgets
whenever you want to hear a full sentence or noun phrase.
ย 
๐Ÿ“ Instructions: Complete each Korean sentence with the best word or phrase. Read the whole sentence aloud, check whether the collocation sounds natural, then open the Answer toggle for the reason.
ย 

Exercise 1 โ€” Reliability Collocations ๐Ÿงญ

A. ์ƒˆ ๊ฒฐ์ œ ์‹œ์Šคํ…œ์€ ์ถœ์‹œ ์ „์— ์‹ ๋ขฐ์„ฑ์„ ์ถฉ๋ถ„ํžˆ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
The new payment system must have its reliability sufficiently verified before launch. ๐Ÿ’ณ
Answer
๊ฒ€์ฆ โ€” ์‹ ๋ขฐ์„ฑ์„ ๊ฒ€์ฆํ•˜๋‹ค means to verify reliability.
B. ์žฅ๊ธฐ๊ฐ„ ์•ˆ์ •์ ์œผ๋กœ ์šด์˜๋˜๋ฉด ์‹œ์Šคํ…œ์˜ ์‹ ๋ขฐ์„ฑ์ด ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์ง„๋‹ค.
If it operates stably for a long time, the systemโ€™s reliability improves. ๐Ÿญ
Answer
๋†’์•„ โ€” ์‹ ๋ขฐ์„ฑ์ด ๋†’์•„์ง€๋‹ค = reliability becomes higher.
C. ๋ฐ˜๋ณต์ ์ธ ์˜ค๋ฅ˜๋Š” ์ž๋™ํ™” ์‹œ์Šคํ…œ์˜ ์‹ ๋ขฐ์„ฑ์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋œจ๋ฆด ์ˆ˜ ์žˆ๋‹ค.
Repeated errors can lower the reliability of an automated system. ๐Ÿงฏ
Answer
๋–จ์–ด โ€” ์‹ ๋ขฐ์„ฑ์„ ๋–จ์–ด๋œจ๋ฆฌ๋‹ค means to reduce reliability.
D. ์™ธ๋ถ€ ๊ฐ์‚ฌ๋Š” ์‹œ์Šคํ…œ ์‹ ๋ขฐ์„ฑ์„ ํ™•๋ณดํ•˜๋Š” ๋ฐ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๊ฐ€ ๋œ๋‹ค.
External audits help secure system reliability. ๐Ÿ“‹
Answer
๋„์›€ โ€” ๋„์›€์ด ๋˜๋‹ค = to be helpful.
E. ์‚ฌ์šฉ์ž๋Š” ๊ฒฐ๊ณผ๊ฐ€ ์ผ๊ด€๋  ๋•Œ ์ž๋™ํ™” ์„œ๋น„์Šค์— ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋ฅผ ๊ฐ–๊ธฐ ์‰ฝ๋‹ค.
Users are likely to have trust in an automated service when results are consistent. ๐Ÿ“ฑ
Answer
์‹ ๋ขฐ โ€” Use ์‹ ๋ขฐ for user trust; use ์‹ ๋ขฐ์„ฑ for system reliability.
F. ๊ธฐ์ˆ  ๊ธฐ์—…์€ ์‹ ๋ขฐ์„ฑ์— ๋Œ€ํ•œ ์˜๋ฌธ์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๊ธฐ ์œ„ํ•ด ์ž๋ฃŒ๋ฅผ ๊ณต๊ฐœํ–ˆ๋‹ค.
The tech company disclosed data to resolve doubts about reliability. ๐Ÿ—‚๏ธ
Answer
ํ•ด์†Œํ•˜ โ€” ์˜๋ฌธ์„ ํ•ด์†Œํ•˜๋‹ค = to resolve doubts.
G. ์‹ ๋ขฐ์„ฑ์ด ๋‚ฎ์€ ๋ชจ๋ธ์€ ์ค‘์š”ํ•œ ๊ฒฐ์ •์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๊ธฐ ์–ด๋ ต๋‹ค.
A model with low reliability is difficult to use for important decisions. โš–๏ธ
Answer
๋งก๊ธฐ โ€” ๊ฒฐ์ •์„ ๋งก๊ธฐ๋‹ค means to entrust a decision.
H. ์šด์˜ ๊ธฐ๋ก์€ ์‹œ์Šคํ…œ์ด ์•ˆ์ •์ ์œผ๋กœ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ–ˆ๋Š”์ง€ ๋ณด์—ฌ ์ค€๋‹ค.
Operation records show whether the system operated stably. ๐Ÿ“Š
Answer
์ž‘๋™ โ€” ์ž‘๋™ํ•˜๋‹ค = to operate/function.
I. ์‹ ๋ขฐ์„ฑ์„ ๋†’์ด๋ ค๋ฉด ๊ฒฐ๊ณผ๋ฟ ์•„๋‹ˆ๋ผ ํŒ๋‹จ ๊ณผ์ •๋„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
To improve reliability, not only results but also the decision process must be reviewed. ๐Ÿ”Ž
Answer
๊ฒ€ํ†  โ€” ๊ฒ€ํ† ํ•˜๋‹ค = to review/examine.
J. ์ž๋™ํ™” ์‹œ์Šคํ…œ์˜ ์‹ ๋ขฐ์„ฑ์€ ํŽธ๋ฆฌํ•จ๋ณด๋‹ค ๋จผ์ € ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋˜์–ด์•ผ ํ•œ๋‹ค.
The reliability of automated systems must be considered before convenience. ๐Ÿงญ
Answer
๊ณ ๋ ค โ€” ๊ณ ๋ ค๋˜๋‹ค = to be considered.
๐Ÿ“š
Vocabulary Box
  • ์‹ ๋ขฐ์„ฑ โ†’ reliability
  • ๊ฒ€์ฆํ•˜๋‹ค โ†’ to verify
  • ๋†’์•„์ง€๋‹ค โ†’ to become higher
  • ๋–จ์–ด๋œจ๋ฆฌ๋‹ค โ†’ to lower
  • ๋„์›€ โ†’ help
  • ์‹ ๋ขฐ โ†’ trust
  • ํ•ด์†Œํ•˜๋‹ค โ†’ to resolve
  • ๋งก๊ธฐ๋‹ค โ†’ to entrust
  • ์ž‘๋™ํ•˜๋‹ค โ†’ to operate
  • ๊ฒ€ํ† ํ•˜๋‹ค โ†’ to review
  • ๊ณ ๋ ค๋˜๋‹ค โ†’ to be considered
Tip ๐Ÿ’ก Reliability sentences love verbs like ๊ฒ€์ฆํ•˜๋‹ค, ํ™•๋ณดํ•˜๋‹ค, ๋†’์ด๋‹ค, and ๊ณ ๋ คํ•˜๋‹ค.

Exercise 2 โ€” Algorithms, Criteria, and Transparency ๐Ÿ”

A. ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ํŒ๋‹จ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์ด ๊ณต๊ฐœ๋˜์ง€ ์•Š์œผ๋ฉด ์‚ฌ์šฉ์ž๋Š” ๊ฒฐ๊ณผ๋ฅผ ๋‚ฉ๋“ํ•˜๊ธฐ ์–ด๋ ต๋‹ค.
If the algorithmโ€™s decision criteria are not disclosed, users have difficulty accepting the result. ๐Ÿง 
Answer
๊ธฐ์ค€ โ€” ํŒ๋‹จ ๊ธฐ์ค€ = decision criteria.
B. ์„ค๋ช… ๊ฐ€๋Šฅํ•œ ์•Œ๊ณ ๋ฆฌ์ฆ˜์€ ์˜์‚ฌ ๊ฒฐ์ • ๊ณผ์ •์˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ๋†’์ธ๋‹ค.
Explainable algorithms increase transparency in decision-making. ๐ŸชŸ
Answer
ํˆฌ๋ช…์„ฑ โ€” Transparency lets users understand how results were produced.
C. ์ž๋ฃŒ๊ฐ€ ๋ถ€์กฑํ•˜๋ฉด ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ํŒ๋‹จ์ด ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋  ์ˆ˜ ์žˆ๋‹ค.
If data is insufficient, the algorithmโ€™s judgment may be distorted. ๐Ÿงฉ
Answer
์™œ๊ณก โ€” ์™œ๊ณก๋˜๋‹ค = to be distorted.
D. ๋ฐ์ดํ„ฐ ํŽธํ–ฅ์€ ์ž๋™ํ™”๋œ ์ถ”์ฒœ ๊ฒฐ๊ณผ์— ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์„ ์ค„ ์ˆ˜ ์žˆ๋‹ค.
Data bias can influence automated recommendation results. ๐Ÿ“ฑ
Answer
์˜ํ–ฅ โ€” ์˜ํ–ฅ์„ ์ฃผ๋‹ค = to affect/influence.
E. ๊ฐœ๋ฐœ์ž๋Š” ์•Œ๊ณ ๋ฆฌ์ฆ˜์ด ์–ด๋–ค ๋ฐฉ์‹์œผ๋กœ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•˜๋Š”์ง€ ์„ค๋ช…ํ•ด์•ผ ํ•œ๋‹ค.
Developers must explain how the algorithm operates. โš™๏ธ
Answer
์ž‘๋™ โ€” ์ž‘๋™ํ•˜๋‹ค = to operate/function.
F. ํŒ๋‹จ ๊ธฐ์ค€์ด ์ž์ฃผ ๋ฐ”๋€Œ๋ฉด ๊ฒฐ๊ณผ์˜ ์ผ๊ด€์„ฑ์ด ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์งˆ ์ˆ˜ ์žˆ๋‹ค.
If decision criteria change often, result consistency can become lower. ๐Ÿ“‰
Answer
๋‚ฎ์•„ โ€” ์ผ๊ด€์„ฑ์ด ๋‚ฎ์•„์ง€๋‹ค = consistency decreases.
G. ๊ณต์ •ํ•œ ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ๋งŒ๋“ค๋ ค๋ฉด ๋ฐ์ดํ„ฐ์˜ ํŽธํ–ฅ์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์•ผ ํ•œ๋‹ค.
To make a fair algorithm, data bias must be reduced. โš–๏ธ
Answer
์ค„์—ฌ โ€” ํŽธํ–ฅ์„ ์ค„์ด๋‹ค = reduce bias.
H. ์‚ฌ์šฉ์ž๊ฐ€ ๊ฒฐ๊ณผ๋ฅผ ์ดํ•ดํ•  ์ˆ˜ ์žˆ๋„๋ก ํŒ๋‹จ ๊ณผ์ •์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•˜๊ฒŒ ์ œ์‹œํ–ˆ๋‹ค.
The decision process was presented transparently so users could understand the result. ๐Ÿ”ฆ
Answer
ํˆฌ๋ช… โ€” ํˆฌ๋ช…ํ•˜๊ฒŒ ์ œ์‹œํ•˜๋‹ค = present transparently.
I. ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ํŽธํ–ฅ์„ ๋ฐœ๊ฒฌํ•˜๋ ค๋ฉด ๊ฒฐ๊ณผ๋ฅผ ๊พธ์ค€ํžˆ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
To find algorithmic bias, results must be monitored steadily. ๐Ÿ‘€
Answer
๊ฐ์‹œ โ€” ๊ฐ์‹œํ•˜๋‹ค = to monitor.
J. ๋ถˆํˆฌ๋ช…ํ•œ ๋ชจ๋ธ์€ ์ฑ…์ž„ ์žˆ๋Š” ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋ฅผ ์–ด๋ ต๊ฒŒ ๋งŒ๋“ ๋‹ค.
An opaque model makes responsible operation difficult. ๐Ÿ—๏ธ
Answer
์šด์˜ โ€” ์šด์˜ = operation/management of a system.
๐Ÿ“š
Vocabulary Box
  • ์•Œ๊ณ ๋ฆฌ์ฆ˜ โ†’ algorithm
  • ํŒ๋‹จ ๊ธฐ์ค€ โ†’ decision criteria
  • ๊ณต๊ฐœ๋˜๋‹ค โ†’ to be disclosed
  • ๋‚ฉ๋“ํ•˜๋‹ค โ†’ to accept/understand
  • ํˆฌ๋ช…์„ฑ โ†’ transparency
  • ์™œ๊ณก๋˜๋‹ค โ†’ to be distorted
  • ํŽธํ–ฅ โ†’ bias
  • ์˜ํ–ฅ์„ ์ฃผ๋‹ค โ†’ to influence
  • ์ž‘๋™ํ•˜๋‹ค โ†’ to operate
  • ์ผ๊ด€์„ฑ โ†’ consistency
  • ๊ฐ์‹œํ•˜๋‹ค โ†’ to monitor
  • ์šด์˜ โ†’ operation
Tip ๐Ÿ’ก For algorithm discussion, pair ํŒ๋‹จ ๊ธฐ์ค€ with ๊ณต๊ฐœํ•˜๋‹ค, and ํŽธํ–ฅ with ์ค„์ด๋‹ค or ๋ฐœ๊ฒฌํ•˜๋‹ค.

Exercise 3 โ€” Error Risk and Safety ๐Ÿงฏ

A. ์ž๋™ํ™”๋˜์–ด๋„ ์˜ค๋ฅ˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์€ ์™„์ „ํžˆ ์‚ฌ๋ผ์ง€์ง€ ์•Š๋Š”๋‹ค.
Even when automated, the possibility of error does not completely disappear. ๐Ÿงฏ
Answer
๊ฐ€๋Šฅ์„ฑ โ€” ์˜ค๋ฅ˜ ๊ฐ€๋Šฅ์„ฑ = possibility of error.
B. ์ƒˆ ์žฅ์น˜๋ฅผ ๋ณ‘์›์— ๋„์ž…ํ•˜๊ธฐ ์ „์— ์•ˆ์ „์„ฑ์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
Before introducing a new device into a hospital, its safety must be verified. ๐Ÿฅ
Answer
๊ฒ€์ฆ โ€” ์•ˆ์ „์„ฑ์„ ๊ฒ€์ฆํ•˜๋‹ค is a core technical collocation.
C. ์‚ฌ์ „ ์ ๊ฒ€์€ ์˜ค๋ฅ˜ ๋ฐœ์ƒ ๊ฐ€๋Šฅ์„ฑ์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋Š” ๋ฐ ํšจ๊ณผ์ ์ด๋‹ค.
Pre-checking is effective in reducing the possibility of errors occurring. ๐Ÿงช
Answer
์ค„์ด โ€” ๊ฐ€๋Šฅ์„ฑ์„ ์ค„์ด๋‹ค = reduce possibility.
D. ์•ˆ์ „ ๊ธฐ์ค€์„ ๊ฐ•ํ™”ํ•˜๋ฉด ์‚ฌ๊ณ ๋ฅผ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•  ์ˆ˜ ์žˆ๋‹ค.
If safety standards are strengthened, accidents can be prevented. ๐Ÿšง
Answer
๋ฐฉ์ง€ โ€” ์‚ฌ๊ณ ๋ฅผ ๋ฐฉ์ง€ํ•˜๋‹ค = prevent accidents.
E. ์˜ค๋ฅ˜๊ฐ€ ๋ฐ˜๋ณต๋˜๋ฉด ์‹œ์Šคํ…œ์˜ ์•ˆ์ •์„ฑ ๋ฌธ์ œ๊ฐ€ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋  ์ˆ˜ ์žˆ๋‹ค.
If errors repeat, a system stability problem may be raised. ๐Ÿ“ฃ
Answer
์ œ๊ธฐ โ€” ๋ฌธ์ œ๊ฐ€ ์ œ๊ธฐ๋˜๋‹ค = a problem is raised.
F. ์ž๋™ํ™” ์ฐจ๋Ÿ‰์€ ๋„๋กœ ํ™˜๊ฒฝ์˜ ๋ณ€์ˆ˜๋ฅผ ์ •ํ™•ํžˆ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
Automated vehicles must accurately predict variables in road conditions. ๐Ÿš—
Answer
์˜ˆ์ธก โ€” ์˜ˆ์ธกํ•˜๋‹ค = to predict.
G. ์ถฉ๋ถ„ํ•œ ์‹คํ—˜ ์—†์ด ์„œ๋น„์Šค๋ฅผ ์ถœ์‹œํ•˜๋ฉด ์œ„ํ—˜์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•˜๊ธฐ ์–ด๋ ต๋‹ค.
If a service launches without sufficient testing, risk is difficult to control. ๐Ÿ”ฅ
Answer
ํ†ต์ œ โ€” ์œ„ํ—˜์„ ํ†ต์ œํ•˜๋‹ค = to control risk.
H. ์˜ค๋ฅ˜ ๋ฐœ์ƒ ๊ฐ€๋Šฅ์„ฑ์„ ๋‚ฎ์ถ”๋ ค๋ฉด ์ž…๋ ฅ ์ž๋ฃŒ๋ฅผ ์ •๊ธฐ์ ์œผ๋กœ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
To lower the possibility of errors, input data must be checked regularly. ๐Ÿ“Š
Answer
์ ๊ฒ€ โ€” ์ž๋ฃŒ๋ฅผ ์ ๊ฒ€ํ•˜๋‹ค = check data.
I. ์•ˆ์ „์„ฑ์ด ํ™•๋ณด๋˜์ง€ ์•Š์€ ์‹œ์Šคํ…œ์€ ๊ณต๊ณต ์„œ๋น„์Šค์— ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•˜๊ธฐ ์–ด๋ ต๋‹ค.
A system whose safety is not secured is difficult to apply to public services. ๐Ÿ›๏ธ
Answer
์ ์šฉ โ€” ์„œ๋น„์Šค์— ์ ์šฉํ•˜๋‹ค = apply to a service.
J. ๊ธฐ์ˆ  ๋ฐœ์ „์€ ํ•„์š”ํ•˜์ง€๋งŒ ์•ˆ์ „์„ฑ ํ™•๋ณด๊ฐ€ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋˜์–ด์•ผ ํ•œ๋‹ค.
Technological development is necessary, but securing safety must come first. ๐Ÿ›ก๏ธ
Answer
์šฐ์„  โ€” ์šฐ์„ ๋˜๋‹ค = to take priority.
๐Ÿ“š
Vocabulary Box
  • ์˜ค๋ฅ˜ ๊ฐ€๋Šฅ์„ฑ โ†’ possibility of error
  • ์•ˆ์ „์„ฑ โ†’ safety as a quality
  • ๊ฒ€์ฆํ•˜๋‹ค โ†’ to verify
  • ์‚ฌ์ „ ์ ๊ฒ€ โ†’ pre-check
  • ์ค„์ด๋‹ค โ†’ to reduce
  • ๋ฐฉ์ง€ํ•˜๋‹ค โ†’ to prevent
  • ์ œ๊ธฐ๋˜๋‹ค โ†’ to be raised
  • ์˜ˆ์ธกํ•˜๋‹ค โ†’ to predict
  • ํ†ต์ œํ•˜๋‹ค โ†’ to control
  • ์ ์šฉํ•˜๋‹ค โ†’ to apply
  • ์šฐ์„ ๋˜๋‹ค โ†’ to take priority
Tip ๐Ÿ’ก Safety writing often uses passive-style forms: ๊ฒ€์ฆ๋˜์–ด์•ผ ํ•œ๋‹ค, ํ™•๋ณด๋˜์–ด์•ผ ํ•œ๋‹ค, ์šฐ์„ ๋˜์–ด์•ผ ํ•œ๋‹ค.

Exercise 4 โ€” Responsibility and Accountability ๐Ÿ“Œ

A. ์ž๋™ํ™” ์‹œ์Šคํ…œ์ด ํ”ผํ•ด๋ฅผ ์ผ์œผ์ผฐ์„ ๋•Œ ์ฑ…์ž„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋ฅผ ๋ช…ํ™•ํžˆ ํ•ด์•ผ ํ•œ๋‹ค.
When an automated system causes harm, responsibility must be clarified. ๐Ÿ“Œ
Answer
์†Œ์žฌ โ€” ์ฑ…์ž„ ์†Œ์žฌ = where responsibility lies.
B. ๊ด€๋ฆฌ ๊ธฐ์ค€์ด ์—†์œผ๋ฉด ๋ฌธ์ œ๊ฐ€ ์ƒ๊ฒผ์„ ๋•Œ ์ฑ…์ž„์ด ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์งˆ ์ˆ˜ ์žˆ๋‹ค.
Without management standards, responsibility may become unclear when a problem occurs. ๐ŸŒซ๏ธ
Answer
๋ถˆ๋ถ„๋ช… โ€” ๋ถˆ๋ถ„๋ช…ํ•ด์ง€๋‹ค = become unclear.
C. ๊ธฐ์—…์€ ์ž๋™ํ™” ๋„๊ตฌ์˜ ๊ฒฐ๊ณผ์— ๋Œ€ํ•œ ๊ด€๋ฆฌ ์ฑ…์ž„์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์•ผ ํ•œ๋‹ค.
Companies must bear management responsibility for the results of automated tools. ๐Ÿข
Answer
์ ธ โ€” ์ฑ…์ž„์„ ์ง€๋‹ค = to take/bear responsibility.
D. ์ฑ…์ž„์„ ํšŒํ”ผํ•˜๋ฉด ๊ธฐ์ˆ ์— ๋Œ€ํ•œ ์‚ฌํšŒ์  ์‹ ๋ขฐ๊ฐ€ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์ง„๋‹ค.
If responsibility is avoided, social trust in technology decreases. ๐Ÿ“‰
Answer
๋‚ฎ์•„ โ€” ์‹ ๋ขฐ๊ฐ€ ๋‚ฎ์•„์ง€๋‹ค = trust decreases.
E. ์˜ค๋ฅ˜๊ฐ€ ๋ฐœ์ƒํ•œ ๋’ค์—๋Š” ์›์ธ์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๊ณ  ์žฌ๋ฐœ์„ ๋ง‰์•„์•ผ ํ•œ๋‹ค.
After an error occurs, the cause must be identified and recurrence prevented. ๐Ÿ”Ž
Answer
๋ฐํžˆ โ€” ์›์ธ์„ ๋ฐํžˆ๋‹ค = identify/reveal the cause.
F. ์ฑ…์ž„ ์žˆ๋Š” ์šด์˜์„ ์œ„ํ•ด ๊ฐ๋… ์ ˆ์ฐจ๋ฅผ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ–ˆ๋‹ค.
Supervision procedures were strengthened for responsible operation. ๐Ÿง‘โ€โš–๏ธ
Answer
๊ฐ•ํ™” โ€” ์ ˆ์ฐจ๋ฅผ ๊ฐ•ํ™”ํ•˜๋‹ค = strengthen procedures.
G. ํ”ผํ•ด์ž๊ฐ€ ๊ตฌ์ œ๋ฅผ ๋ฐ›์œผ๋ ค๋ฉด ์ฑ…์ž„ ์ฃผ์ฒด๊ฐ€ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
For victims to receive relief, the responsible party must be clear. ๐Ÿงพ
Answer
๋ช…ํ™• โ€” ๋ช…ํ™•ํ•˜๋‹ค = clear.
H. ์ž๋™ํ™” ๊ฒฐ์ •์— ์‚ฌ๋žŒ์ด ๊ด€์—ฌํ•˜์ง€ ์•Š์•˜๋‹ค๋Š” ์ด์œ ๋กœ ์ฑ…์ž„์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•  ์ˆ˜๋Š” ์—†๋‹ค.
Responsibility cannot be avoided just because no person was involved in the automated decision. ๐Ÿค–
Answer
ํšŒํ”ผ โ€” ์ฑ…์ž„์„ ํšŒํ”ผํ•˜๋‹ค = avoid responsibility.
I. ์ •์ฑ…์€ ๊ธฐ์ˆ ์˜ ํŽธ๋ฆฌํ•จ๊ณผ ์‚ฌํšŒ์  ์ฑ…์ž„์„ ํ•จ๊ป˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
Policy must consider both technological convenience and social responsibility. โš–๏ธ
Answer
๊ณ ๋ ค โ€” ํ•จ๊ป˜ ๊ณ ๋ คํ•˜๋‹ค = consider together.
J. ๋ฌธ์ œ๊ฐ€ ๋ฐ˜๋ณต๋˜๋ฉด ๋‹จ์ˆœํ•œ ์˜ค๋ฅ˜๊ฐ€ ์•„๋‹ˆ๋ผ ๊ตฌ์กฐ์  ๊ฒฐํ•จ์œผ๋กœ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋  ์ˆ˜ ์žˆ๋‹ค.
If problems repeat, they may be viewed as structural defects rather than simple errors. ๐Ÿงฑ
Answer
๊ฐ„์ฃผ โ€” ๊ฐ„์ฃผ๋˜๋‹ค = to be regarded as.
๐Ÿ“š
Vocabulary Box
  • ์ฑ…์ž„ ์†Œ์žฌ โ†’ where responsibility lies
  • ํ”ผํ•ด โ†’ harm/damage
  • ๋ช…ํ™•ํžˆ ํ•˜๋‹ค โ†’ to clarify
  • ๋ถˆ๋ถ„๋ช…ํ•ด์ง€๋‹ค โ†’ to become unclear
  • ์ฑ…์ž„์„ ์ง€๋‹ค โ†’ to bear responsibility
  • ์ฑ…์ž„์„ ํšŒํ”ผํ•˜๋‹ค โ†’ to avoid responsibility
  • ์›์ธ์„ ๋ฐํžˆ๋‹ค โ†’ to identify the cause
  • ๊ฐ•ํ™”ํ•˜๋‹ค โ†’ to strengthen
  • ๋ช…ํ™•ํ•˜๋‹ค โ†’ to be clear
  • ๊ณ ๋ คํ•˜๋‹ค โ†’ to consider
  • ๊ฐ„์ฃผ๋˜๋‹ค โ†’ to be regarded as
Tip ๐Ÿ’ก Accountability writing often asks: ๋ˆ„๊ฐ€ ์ฑ…์ž„์„ ์ง€๋Š”๊ฐ€? (Who bears responsibility?)

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

A. โŒ ์ž๋™ํ™” ์‹œ์Šคํ…œ์€ ํ•ญ์ƒ ๊ฐ๊ด€์ ์ด๋‹ค. โœ… ๋ฐ์ดํ„ฐ๊ฐ€ ํŽธํ–ฅ๋˜๋ฉด ๊ฒฐ๊ณผ๊ฐ€ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋  ์ˆ˜ ์žˆ๋‹ค.
If data is biased, results can be distorted. โš–๏ธ
Answer
์™œ๊ณก โ€” Pitfall check: algorithms are not automatically objective; biased data can distort results.
B. โŒ ํŽธ๋ฆฌํ•˜๋ฉด ์ถฉ๋ถ„ํ•˜๋‹ค. โœ… ํŽธ๋ฆฌํ•จ๋ณด๋‹ค ์‹ ๋ขฐ์„ฑ๊ณผ ์•ˆ์ „์„ฑ์„ ๋จผ์ € ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
Reliability and safety must be considered before convenience. ๐Ÿงญ
Answer
๊ณ ๋ ค โ€” Advanced evaluation balances convenience with reliability and safety.
C. โŒ ์˜ค๋ฅ˜๋Š” ์‚ฌ๋žŒ๋งŒ ๋งŒ๋“ ๋‹ค. โœ… ์ž๋™ํ™” ์‹œ์Šคํ…œ์—๋„ ์˜ค๋ฅ˜ ๊ฐ€๋Šฅ์„ฑ์ด ๏ผฟ๏ผฟ๏ผฟ๏ผฟ ์žˆ๋‹ค.
Automated systems can also have the possibility of error. ๐Ÿงฏ
Answer
๋‚จ์•„ โ€” ๊ฐ€๋Šฅ์„ฑ์ด ๋‚จ์•„ ์žˆ๋‹ค = the possibility remains.
D. โŒ ์•Œ๊ณ ๋ฆฌ์ฆ˜์„ ๋ฏฟ์ž. โœ… ์•Œ๊ณ ๋ฆฌ์ฆ˜์˜ ์‹ ๋ขฐ์„ฑ์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•˜์ž.
Letโ€™s verify the reliability of the algorithm. ๐Ÿ”Ž
Answer
๊ฒ€์ฆ โ€” Use verification language, not blind-trust language.
E. โŒ ๋ฌธ์ œ๊ฐ€ ์ƒ๊ธฐ๋ฉด ๋ˆ„๊ตฐ๊ฐ€ ์ฑ…์ž„์ง„๋‹ค. โœ… ์ฑ…์ž„ ์†Œ์žฌ๋ฅผ ์‚ฌ์ „์— ๏ผฟ๏ผฟ๏ผฟ๏ผฟํžˆ ํ•ด์•ผ ํ•œ๋‹ค.
Accountability must be clarified in advance. ๐Ÿ“Œ
Answer
๋ช…ํ™• โ€” ์ฑ…์ž„ ์†Œ์žฌ๋ฅผ ๋ช…ํ™•ํžˆ ํ•˜๋‹ค is the natural phrase.
F. โŒ ์•ˆ์ „๋งŒ ๋ณด๋ฉด ๋œ๋‹ค. โœ… ์•ˆ์ „์„ฑ๊ณผ ์‹ ๋ขฐ์„ฑ์„ ํ•จ๊ป˜ ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
Safety and reliability must be verified together. ๐Ÿ›ก๏ธ
Answer
๊ฒ€์ฆ โ€” Technical assessment should verify both.
G. โŒ ํŒ๋‹จ ๊ณผ์ •์€ ์ˆจ๊ฒจ๋„ ๋œ๋‹ค. โœ… ํŒ๋‹จ ๊ณผ์ •์˜ ํˆฌ๋ช…์„ฑ์„ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์•ผ ํ•œ๋‹ค.
Transparency in the decision process must be increased. ๐ŸชŸ
Answer
๋†’์—ฌ โ€” ํˆฌ๋ช…์„ฑ์„ ๋†’์ด๋‹ค = increase transparency.
H. โŒ ํ•œ ๋ฒˆ ์„ฑ๊ณตํ•˜๋ฉด ๋์ด๋‹ค. โœ… ์šด์˜ ์ค‘์—๋„ ๊ฒฐ๊ณผ๋ฅผ ๊ณ„์† ๏ผฟ๏ผฟ๏ผฟ๏ผฟํ•ด์•ผ ํ•œ๋‹ค.
Even during operation, results must continue to be monitored. ๐Ÿ‘€
Answer
๊ฐ์‹œ โ€” Reliability requires ongoing monitoring, not one-time success.
I. โŒ ์‹ค์ˆ˜์™€ ์˜ค๋ฅ˜๋Š” ๊ฐ™๋‹ค. โœ… ์‹œ์Šคํ…œ ๋ฌธ์ œ์—๋Š” ๋ณดํ†ต ๏ผฟ๏ผฟ๏ผฟ๏ผฟ๋ผ๋Š” ๋ง์„ ์“ด๋‹ค.
For system problems, the word โ€œerrorโ€ is usually used. ๐Ÿงฉ
Answer
์˜ค๋ฅ˜ โ€” ์‹ค์ˆ˜ is more human; ์˜ค๋ฅ˜ fits systems/data/logic.
J. โŒ ์‹ ๋ขฐ์™€ ์‹ ๋ขฐ์„ฑ์€ ๊ฐ™๋‹ค. โœ… ์‹œ์Šคํ…œ์˜ ๊ฐ๊ด€์  ํ’ˆ์งˆ์€ ๏ผฟ๏ผฟ๏ผฟ๏ผฟ์ด๋ผ๊ณ  ํ•œ๋‹ค.
The objective quality of a system is called reliability. ๐Ÿงญ
Answer
์‹ ๋ขฐ์„ฑ โ€” ์‹ ๋ขฐ = trust; ์‹ ๋ขฐ์„ฑ = reliability.
๐Ÿ“š
Vocabulary Box
  • ์™œ๊ณก๋˜๋‹ค โ†’ to be distorted
  • ๊ณ ๋ คํ•˜๋‹ค โ†’ to consider
  • ๋‚จ์•„ ์žˆ๋‹ค โ†’ to remain
  • ๊ฒ€์ฆํ•˜๋‹ค โ†’ to verify
  • ์ฑ…์ž„ ์†Œ์žฌ โ†’ where responsibility lies
  • ๋ช…ํ™•ํžˆ ํ•˜๋‹ค โ†’ to clarify
  • ํˆฌ๋ช…์„ฑ โ†’ transparency
  • ๋†’์ด๋‹ค โ†’ to increase
  • ๊ฐ์‹œํ•˜๋‹ค โ†’ to monitor
  • ์˜ค๋ฅ˜ โ†’ system/data/logical error
  • ์‹ค์ˆ˜ โ†’ human mistake
  • ์‹ ๋ขฐ์„ฑ โ†’ reliability
Tip ๐Ÿ’ก Pitfall patrol: replace vague praise with precise evaluation โ€” ๊ฒ€์ฆํ•˜๋‹ค, ๊ณ ๋ คํ•˜๋‹ค, ๊ฐ์‹œํ•˜๋‹ค, ๋ช…ํ™•ํžˆ ํ•˜๋‹ค.
ย 
๐ŸŽ‰ You finished the reliability circuit! You can now discuss automated systems with the advanced vocabulary of safety, bias, transparency, error risk, and accountability โ€” which is exactly how to sound thoughtful instead of just impressed by shiny machines. ๐Ÿค–๐Ÿ›ก๏ธ