Ethics and bias risk: Which category of AI risk does this situation most directly illustrate?

Systematically unfair outcomes for a demographic group are the defining illustration of ethics and bias risk.

The question

Preparing for AIGP? Take the free 5-min readiness quiz →

1. Misalignment risk, where the model optimizes a proxy goal that diverges from the truly intended objective

   Plausible but wrong: misalignment concerns objective-proxy divergence, not unequal treatment of groups; nothing here indicates a mis-specified goal.
2. Ethics and bias risk, where the system produces systematically unfair outcomes for particular groups ✓

   Correct: systematically lower scores for one group despite equal qualifications is the textbook example of ethics and bias risk.
3. Complexity and scalability risk, where harms multiply as the system is deployed across contexts

   Plausible but wrong: this describes harms compounding at scale, whereas the scenario centers on unfair scoring of a group.
4. Data-dependency risk, where poor training-data quality degrades the reliability of predictions

   Almost right: bad data can be a cause, but the risk being illustrated by the unfair outcome itself is bias, not general reliability loss.

How to remember it

Systematically unfair outcomes for a demographic group are the defining illustration of ethics and bias risk.

More Understanding the Foundations of AI Governance questions

• It produces new content such as text: In common AI terminology, what most distinctly characterizes 'generative →
• Probabilistic outputs and opacity: Which pair of AI characteristics best explains this governance challenge? →
• Transparency and explainability: Which responsible-AI principle is the bank primarily operationalizing? →
• All 337 Understanding the Foundations of AI Governance questions →