Skip to content

CS Professional · Artificial Intelligence, Data Analytics and Cyber Security - Laws and Practice · Artificial Intelligence - Introduction and Basics

A lender in Pune uses a machine-learning model to screen loan applications. Audit finds that the model rejects applicants from certain PIN codes far more often, because past approval data reflected earlier discriminatory lending. Which AI risk does this best illustrate?

This illustrates algorithmic bias arising from skewed training data. The model learned from historical approvals that reflected past discrimination, so it repeats unfair outcomes against certain PIN codes. It is a fairness problem in the data, not overfitting, an adversarial attack or data theft.

  1. AAlgorithmic bias arising from skewed training dataCorrect
  2. BModel overfitting caused by too many training epochs
  3. CAdversarial attack through manipulated input images
  4. DData exfiltration through a compromised API key

Explanation

The model learned patterns from historical data that already carried discriminatory outcomes, so it reproduces and may amplify them. That is algorithmic (data) bias. Overfitting concerns poor generalisation to new data, not inherited unfairness; the other options describe security attacks that are not described in the facts.

Did you get it right without looking?

One question tells you little. A timed set on Artificial Intelligence - Introduction and Basics shows your real accuracy, how long you take and where you lose marks.

Practise Artificial Intelligence - Introduction and BasicsFree diagnostic mock

More Artificial Intelligence - Introduction and Basics questions