Professional Standards & Ethics Flashcards
7 cards from real AML practice questions. Tap to flip, then mark Knew It or Still Learning — missed cards come back until you master them.
Read the first 7 Professional Standards & Ethics flashcards as text
Which of the following scenarios BEST exemplifies 'automation bias' as an ethical risk in ML deployment?
Answer: A physician accepts an AI diagnosis without applying independent clinical judgment
Automation bias occurs when humans over-trust automated systems and fail to apply independent critical judgment, increasing risk of undetected errors.
Under the ACM Code of Ethics, a computing professional's HIGHEST obligation when their employer instructs them to act unethically is to:
Answer: Prioritize public good and refuse directives that violate ethical principles
The ACM Code of Ethics establishes that public good and ethical principles take precedence over organizational loyalty and instructions.
A company trains a large language model on scraped web data that includes copyrighted text. Which ethical and legal risk does this PRIMARILY raise?
Answer: Intellectual property infringement and potential copyright liability
Training on copyrighted material without license or fair use justification raises direct intellectual property infringement risks.
What is the MOST important characteristic distinguishing an 'ethical review board' (or IRB) from a standard legal compliance review for ML projects?
Answer: Ethical review proactively evaluates moral risks to individuals beyond minimum legal requirements
Ethical review goes beyond legal minimums to consider broader harms, rights, and societal impacts that law may not yet address.
A feature importance analysis reveals that a lending model heavily weights 'neighborhood' as a proxy feature. This MOST directly suggests the presence of:
Answer: Proxy discrimination using a legally protected attribute
Neighborhood is a well-known proxy for race and national origin; heavy reliance on it constitutes proxy discrimination even if protected attributes are excluded.
When a model's decision is challenged in a high-stakes context, the practitioner's ethical duty of 'explainability' MOST directly serves which goal?
Answer: Enabling affected individuals to understand and contest the decision
Explainability in high-stakes contexts primarily serves the rights of individuals to understand decisions affecting them and to seek meaningful redress.
An ML team is asked to build a predictive policing model for a US city. Which ethical consideration MOST strongly argues against deployment?
Answer: Such models risk amplifying historical racial bias in policing and producing self-fulfilling enforcement loops
Predictive policing systems trained on biased historical arrest data risk entrenching systemic racial disparities and creating feedback loops where predictions become self-fulfilling.