IT Specialist AI Exam 1.4-1.6

Front

Which of the following best describes the purpose of internal access levels in an AI system? To allow unrestricted model training, To maximize the amount of data users can access, To ensure only authorized users can access certain data or system functions, To improve system performance

Back

To ensure only authorized users can access certain data or system functions

Front

What is an example of an adversarial attack on an AI model?

Back

Adding slightly altered inputs to trick a model into making incorrect predictions.

Front

When evaluating appropriate use of AI, which question is most important? How many developers built the model, How much storage the model requires, Whether the model trains quickly, Whether the model might harm or mispredict for specific user groups

Back

Whether the model might harm or mispredict for specific user groups

Front

Setting guidelines for data gathering primarily ensures that:

Back

Data is collected ethically, with transparency and user protection

Front

Out-of-context AI results are dangerous because:

Back

They may be interpreted incorrectly and lead to harmful decisions

Front

Why is transparency about the purpose of data collection important?

Back

It helps users understand how and why their data is used

Front

Determining who should see AI results relates to which concept?

Back

Access control and stakeholder permissions

Front

Industry-specific legal requirements must be reviewed because:

Back

Legal rules vary across sectors, especially in areas like health, finance, and education

Front

What is one reason why AI outputs must include confidence levels or uncertainty explanations?

Back

They help users gauge reliability and make informed decisions about using results.

Front

Which of the following is an example of how an AI system might unintentionally harm a specific user group? A spellchecker suggests synonyms for commonly used words, A facial recognition system has higher error rates for individuals with darker skin tones, leading to misidentification, A calendar app adds customizable color themes, A weather app displays hourly forecasts in a new layout

Back

A facial recognition system has higher error rates for individuals with darker skin tones, leading to misidentification