CT-AI practice question 15 of 15

A beer company is trying to understand how much recognition its logo has in the market. It plans to do that by monitoring images on various social…

A beer company is trying to understand how much recognition its logo has in the market. It plans to do that by monitoring images on various social media platforms using a pre-trained neural network for logo detection. This particular model has been trained by looking for words, as well as matching colors on social media images. The company logo has a big word across the middle with a bold blue and magenta border. Which associated risk is most likely to occur when using this pre-trained model?

Answer options

A: There is no risk, as the model has already been trained

B: Insufficient function; the model was not trained to check for colors or words

C: Improper data preparation

D: Inherited bias: the model could have inherited unknown defects Correct answer

Correct Answer:

D. Inherited bias: the model could have inherited unknown defects

Explanation

Using a pre-trained model, a form of transfer learning, introduces the risk of inheriting any flaws from the original model's development. Since the new user typically has no access to the original training data or process, the model may contain unknown biases, defects, or vulnerabilities. These "inherited" issues can lead to unexpected and incorrect behavior when the model is applied to a new domain, even if the high-level function seems appropriate.

Why Incorrect

A. There is no risk, as the model has already been trained: This is incorrect. Using any third-party component, especially a complex one like an AI model, always introduces risks.
B. Insufficient function; the model was not trained to check for colors or words: The question states the model was trained on these features, so this option contradicts the premise.
C. Improper data preparation: This risk is associated with training a new model, not with using an existing, pre-trained one where the data preparation has already been done.

References

ISTQB Certified Tester AI Testing (CT-AI) Syllabus Version 1.0

Section 4.3.2

"Testing Pre-Trained Models

" states

"the test team has to deal with the risk of inheriting unknown defects."

Raji

I. D.

& Buolamwini

J. (2019). "Actionable Auditing: Investigating the Impact of Publicly Naming Biased Performance Results of Commercial AI Products." Proceedings of the 2019 AAAI/ACM Conference on AI

Ethics

and Society

pp. 429-435. This paper demonstrates how commercial

pre-trained models can inherit and propagate biases. DOI: https://doi.org/10.1145/3306618.3314244

Keep practicing

Take the free CT-AI practice test

Ten exam-style questions with answers and explanations, plus the exam facts and study guides.

Start the free practice test