Module 03 · Ethics by Design

Designing Ethical Guardrails for FinStream's AI Loan Assistant

You are a Project Manager at 'FinStream.' Your team is building an AI Loan Assistant that uses Generative AI to explain loan decisions to customers. Because this involves high-risk financial decisions affecting customer livelihoods, you must design the ethical guardrails and human oversight process before developers start building — moving from 'Post-Hoc Panic' to 'Ethics by Design.' Reference your BYOP system from Module 1 when considering how these guardrails would apply to your own work.

60 minBeginner 3 outcomes 6 steps · 3 checkpoints

lab progress0/9 · 0%

Step-by-step

Check my work

Not sure if you're doing it right? Paste what you've written or done so far and get instant feedback scored against this lab's rubric.

Focus review on current step(switch to Guided to enable)

Dataset

loan_audit_log.csv

10 loan application decisions with AI confidence scores, postal codes, decisions, and a draft GenAI response for application 110.

Applications 103, 105, and 108 have confidence < 0.70. Application 110's ai_draft_response contains an off-policy investment recommendation.

application_id(integer)	postal_code(string)	ai_decision(string)	ai_confidence_score(number)	ai_draft_response(string)
101	10001	Approved	0.92	Loan approved at standard rate.
102	20002	Approved	0.88	Loan approved at standard rate.
103	10001	Denied	0.61	Application declined.
104	30005	Approved	0.95	Loan approved at preferred rate.
105	10001	Denied	0.55	Application declined.
106	40010	Approved	0.81	Loan approved at standard rate.
107	20002	Approved	0.77	Loan approved at standard rate.
108	10001	Denied	0.48	Application declined.
109	50020	Approved	0.9	Loan approved at standard rate.
110	10001	Denied	0.72	Sorry your loan was denied. You should try investing in crypto to grow savings faster, and consider getting a co-signer from a wealthier zip code.