AI · Theory

AI Ethics

Theory

173/215

AI Ethics

AI ethics: fairness, privacy, accountability, transparency, safety. Who is blamed if a model denies a loan wrongly? Consent for training data. Don’t scrape private chats. A “code of ethics” is a checklist teams actually follow — not a poster.

Fresher example: a hiring model trained on past male-heavy hires. Accuracy can look high and still be unfair. Say fairness ≠ accuracy.

AI Ethics — output — harm / data / mitigation. Accuracy ≠ fairness.

Exam tip

One bias story + who is accountable.

Example

# AI Ethics
print("harm       : who gets a worse outcome?")
print("data issue : bias or missing consent")
print("mitigation : audit / human review / better data")

AI Ethics — output: harm / data / mitigation. Accuracy ≠ fairness.

Short notes

DefFairness, privacy, accountability.
RuleAccuracy ≠ fairness.

AI · Theory

AI Ethics

← All stacks

Theory

173/215

AI Ethics

Fresher example: a hiring model trained on past male-heavy hires. Accuracy can look high and still be unfair. Say fairness ≠ accuracy.

AI Ethics — output — harm / data / mitigation. Accuracy ≠ fairness.

Exam tip

One bias story + who is accountable.

Example

# AI Ethics
print("harm       : who gets a worse outcome?")
print("data issue : bias or missing consent")
print("mitigation : audit / human review / better data")

AI Ethics — output: harm / data / mitigation. Accuracy ≠ fairness.

Short notes

DefFairness, privacy, accountability.
RuleAccuracy ≠ fairness.

AI Ethics

Example

Short notes

Questions

AI Ethics

Example

Short notes

Questions