BiasWesley

Question 1

What type of bias occurs when a dataset favors a specific viewpoint, like only including customer queries about certain insurance policies?

Answer 1

Confirmation bias

Question 2

If a chatbot is trained on data from several years ago and struggles with current customer queries, what type of bias is likely present?

Answer 2

Historical bias

Question 3

What type of bias arises when the labels applied to customer query data are subjective, inaccurate, or too generic?

Answer 3

Labelling bias

Question 4

A chatbot trained primarily on formal written language struggles to understand informal language. What kind of bias is this?

Answer 4

Linguistic bias

Question 5

If a training dataset for a chatbot only includes customer queries from one demographic, what type of bias is present?

Answer 5

Sampling bias

Question 6

What type of bias occurs when training data is chosen based on specific criteria instead of being randomly selected?

Answer 6

Selection bias

Question 7

Define bias in the context of chatbot datasets.

Answer 7

A systematic error that can lead to unfair or inaccurate predictions.

Question 8

Why is it important to address biases in chatbot datasets?

Answer 8

To ensure fairness, accuracy, and avoid perpetuating existing societal biases.

Question 9

Give an example of how historical bias could negatively impact a chatbot’s performance.

Answer 9

The chatbot may not accurately understand or respond to current slang, trending topics, or changes in customer needs.

Question 10

What are some general strategies for mitigating bias in chatbot datasets?

Answer 10

Use diverse datasets, carefully review and correct labels, and regularly audit the chatbot’s performance for bias.