Github Copilot · Free Practice Question Medium
Question 39
You are using GitHub Copilot to generate code for a healthcare system handling sensitive data like patient records. The code needs to be reliable and unbiased. Aware that Copilot's suggestions come from a large training corpus, you're concerned about its limitations in this critical context. How can you address these limitations when developing your system?
-
A
Manually review all code suggestions for accuracy, keeping in mind that the training data may not reflect the latest medical standards or regulations.
-
B
Disable Copilot’s suggestions in regulated industries such as healthcare to avoid any potential liability due to the unknown quality of the training data.
-
C
Assume that because Copilot is trained on a large dataset, it will generate highly accurate code without the need for additional review.
-
D
Rely on Copilot’s suggestions for simple, non-critical code, but use manual coding methods for complex or sensitive parts of the application.
Reveal correct answer
Correct answer: A
A.
Copilot’s training data may not always include up-to-date or industry-specific guidelines, especially in rapidly evolving fields like healthcare. Therefore, it’s crucial to review its suggestions carefully, particularly when sensitive or regulated data is involved. Manually ensuring that the code complies with the latest standards and regulations is the safest approach to mitigate the limitations of generative AI models.
B.
Disabling Copilot entirely in such industries is not necessary. Instead, responsible use of the tool—through careful review and consideration of its limitations—allows developers to benefit from AI without compromising legal or ethical standards.
C.
The size of the training dataset does not guarantee accuracy or relevance. Copilot’s training data may contain outdated or biased information, which means the generated code requires human oversight, especially in critical fields like healthcare.
D.
While this strategy might reduce risk in sensitive areas, even simple code suggestions can have subtle errors or biases that affect the overall system. A responsible approach requires review of all suggestions, regardless of complexity.
Discussion
Think the marked answer is wrong, or have a better explanation? Share it below — comments appear after review.
