What role do ethics play in your work as a data scientist?

Example answer

Ethics affects whether a data project should proceed, what data it should use and how its results affect people. I start by identifying the intended benefit, the affected groups and plausible harms, including people who are not direct users. Technical feasibility or permission to access a dataset is not sufficient justification for every proposed use.

I check the data's origin, permitted purpose, necessity, access and retention with the responsible owners. I limit collection and disclosure to what the task needs and involve privacy, security or other qualified reviewers where their judgment is required. Removing a name does not by itself establish that a record cannot identify someone.

For an analysis or model, I examine label quality, coverage, relevant group outcomes and uncertainty. I document limitations and tradeoffs, including conflicts between metrics. I consider how people can question or correct an important result and who owns monitoring and response after release. If a consequential risk remains unresolved, I recommend a narrower use, further review or stopping the work rather than certify it as unbiased.

FAIR means Findable, Accessible, Interoperable and Reusable: it concerns data stewardship and does not by itself establish ethical use, privacy or algorithmic fairness. The voluntary NIST AI Risk Management Framework can help structure AI risk work; following a framework is not a universal legal-compliance certificate.

Follow-up to practice

Give a truthful example of a tradeoff, who made the accountable decision and what limitation you documented. Keep confidential details out.

Sources: FAIR principles; NIST AI Risk Management Framework; ICO on indirect identification.